[Next] [Up/Previous] [Mathematics] [Other] [Home]

An Aside on the Slot Machine

Note: to supply some context, images of the slot machines mentioned, taken from old advertisements for them, have been included where possible. However, much clearer images, in full color even, are easily available with a quick web search if one is interested.

Some books on gambling illustrated the following as a typical slot machine arrangement:

(%) Cherries     5   7   3
(O) Oranges      4   1  10
(@) Plums        6   1   4
(A) Bells        1   9   1
(U) Horseshoes   2   1   1
(=) Bars         2   1   1

perhaps with payouts like these:

%..             3
%%.             5
%%%            11
OOO, OO=       11
@@@, @@=       13
AAA, AA=       18
UUU            18
===           100

if it were, as I had recently come to suspect, a Jennings Standard Chief Bell or Jennings Export Chief, which indeed had those payouts.

This is a more modern arrangement, of the general style favored in the postwar era. Lemons are completely removed from the machine; instead, the horseshoe is added. Some accounts of this arrangement instead refer to the added symbol as a melon. Many modern machines have a second jackpot symbol; while earlier machines with a jackpot symbol other than the bar (and on some of them, three bars only won 20 coins rather than a jackpot) used a gold coin symbol for a "gold award", the melon, as well as a red number 7, were common added jackpot symbols, so I had originally expected three horseshoes to win a jackpot instead of just 18 coins.

And now even a single cherry on the first reel wins. Note, however, that there were many different payout arrangements used in the postwar era; no one arrangement was nearly universal, as was the case with the prewar era. And often, both pre-war and postware, even slot machines with the same model name had different reel strips and corresponding payout discs available for them.

The melon as an extra jackpot symbol appears to have been introduced by Mills on their Melon Bell machine; they've also used a 7 in this role instead. The melon was later also found on Bally and Sega machines.

Here, a greater effort has been taken to limit the number of times a winning combination occurs. There is only one bell on the first and third reels, so the second reel can safely have nine bells on it, and so there is only one orange, and one plum, on the second reel.

We can note that if every possible combination of the reels occurs exactly once in 8000 plays, then the machine takes in 8000 coins, it will give out the number of coins calculated as follows:

Reel           Combinations  Value   Amount
 1    2    3                         Paid

 5 * 13 * 20 =         1300  *   3 =  3900
 5 *  7 * 17 =          595  *   5 =  2975
 5 *  7 *  3 =          105  *  11 =  1155
 4 *  1 * 11 =           44  *  11 =   484
 6 *  1 *  5 =           30  *  13 =   390
 1 *  9 *  2 =           18  *  18 =   324
 2 *  1 *  1 =            2  *  18 =    36
 2 *  1 *  1 =            2  * 100 =   200  

and, of course, this can't be right, and so that arrangement of symbols must belong to the reel strips of a machine with smaller payouts in some respect.

Changing the payout for one cherry to only two coins would reduce the total payout to 8064 coins, and so some additional changes, such as paying 3 coins for two cherries and 5 coins for three cherries would be enough to make the machine at least slightly profitable. As the book from which I learned of this machine mentioned the Jennings Lucky Bell was an earlier, almost identical machine, perhaps it was the one with this schedule of symbols, and lower payouts such as those.

And only a slight change would be needed to make the higher payouts profitable; one of the simplest would be to replace some of the cherries on the first reel by additional horseshoes.

So the basic idea behind a slot machine is that the chances of a winning combination are reduced by having very unequal numbers of symbols on the different wheels.

This particular arrangement resembles some used on Jennings slot machines known to me. For example, the V-12-70 set of reel strips, which does not include an extra jackpot symbol, has this arrangement:

(%) Cherries     3   7   3
(O) Oranges      7   1  10
(@) Plums        6   1   5
(A) Bells        1  10   1
(=) Bars         3   1   1

And their payouts would be:

Reel           Combinations  Value   Amount
 1    2    3                         Paid

 3 * 13 * 20 =          580  *   3 =  1740
 3 *  7 * 17 =          357  *   5 =  1785
 5 *  7 *  3 =          105  *  11 =  1155
 7 *  1 * 11 =           77  *  11 =   847
 6 *  1 *  6 =           36  *  13 =   468
 1 * 10 *  2 =           20  *  18 =   360
 3 *  1 *  1 =            3  * 100 =   300  

for a profit of 16.8125%.

And then there's V-12-88, which has this arrangement:

(%) Cherries     2   8   3
(O) Oranges      8   1  10
(@) Plums        6   3   5
(A) Bells        2   8   1
(=) Bars         2   1   1

This principle could be taken to extremes, as this example illustrates:

(*) Lemons       2   1   5
(%) Cherries     5   2   1
(O) Oranges      1   1  10
(@) Plums        1   5   2
(A) Bells        1  10   1
(=) Bars        10   1   1

With 20 symbols on each wheel, three of the same symbol, one on each reel, can only come up 10 times for any one symbol, for a total of 60 times for all six different symbols. So that could pay out a jackpot of 100 coins in each case, giving a total payout of 6000 coins out of 8000, or a payout percentage of 75%.

While there were actual slot machines with 9 or 10 occurences of the same symbol on one reel, that means that one often sees the same symbol twice among the three visible symbols on any given reel. If one were to restrict oneself to using any given symbol at most six times on a given reel, so that this need never occur, could one still design a machine that could be profitable paying only jackpots?

Since a 50 coin payout still counts as a jackpot, it is possible:

(O) Oranges     6   4   1
(@) Plums       1   6   4
(A) Bells       4   1   6
(=) Bars        6   1   2
(C) Melons      2   6   1
(7) Sevens      1   2   6

If three bars, three melons, and three sevens each paid 100, then each symbol would pay 1200 coins, since in each case three of that symbol could occur 12 times. Each of the first three symbols occurs three in a row twice as often, so if the payout were 50, half as much, then we have six symbols each paying 1200 coins, for a total payout of 7200 out of 8000, or 90%. Of course, a jackpot-only machine should really have a lower percentage than that, unlike one with many small payouts.

There have been actual slot machines made that only pay jackpots. Most of them have numerous blank positions on the reels, but in the book Slot Machines: A Pictorial History of the First 100 Years, there is pictured a Mills hi-top slot machine modified by the Nugget casino in Sparks, Nevada (not the Golden Nugget casino in Las Vegas) so that all the winning combinations paid jackpots without resorting to that expedient; its payout schedule is shown at right.

I don't know how the symbols were distributed on its reels, but one possibility might be something like this:

(O) Oranges     6   5   1
(@) Plums       1   6   3
(A) Bells       3   1   6
(=) Bars        6   1   3
(C) Melons      3   6   1
(7) Sevens      1   1   6

which would lead to the following payout:

6 * 5 * 1 =  30    *  60 = 1800
1 * 6 * 3 =  18    *  60 = 1080
3 * 1 * 6 =  18    *  60 = 1080
6 * 1 * 3 =  18    *  60 = 1080
3 * 6 * 1 =  18    *  60 = 1080
1 * 1 * 6 =   6    * 200 = 1200

which would lead to a house advantage of 8.5%; just by allowing up to 7 of the same symbol on a reel, so that the same symbol would occasionally be visible twice, that could easily be considerably increased:

7 * 3 * 1 =  21    *  60 = 1260
1 * 7 * 2 =  14    *  60 =  840
2 * 1 * 7 =  14    *  60 =  840
7 * 1 * 2 =  14    *  60 =  840
2 * 7 * 1 =  14    *  60 =  840
1 * 1 * 7 =   7    * 200 = 1400

for a house percentage of 37.25%.

The following diagram shows

how, as the number of times that we want the same symbol to appear on a reel with 20 positions increases, if we try to keep them as evenly spaced as possible, the number of positions in which two of the same symbol are visible in a normal 3-line window will increase. (Of course, the circles filled with red represent the locations of the symbol that is being repeated, and the circles filled with light blue-green are the locations which, when they come up on the center line, will be flanked by two of that symbol, one above and one below.)

As for those with blank positions on the reels, sometimes they had positions that were actually blank, as on the Seeburg 3-Line Stars and Bars and many others, and in at least one case, a Pace Deluxe Cherry Bell, which closely resembles the Pace Comet, modified by another casino, the Harrah's Club casino, had each reel bearing one word from the phrase "Pays Jackpots Only" in the positions that would not form part of any prize.

Starting from the Beginning

The following image illustrates some of the ways that slot machines have changed over the years.

The image above is from Wikimedia Commons, licensed under the GNU Free Documentation License, and is thus available for your use under the same terms.

Its author is Nazox; the image here is resized and cropped from the original.

In the top row, the first set of payouts shown are those of the first slot machine (but not with the first set of reel strips used with it), as invented by Charles Fey. A photograph of the specimen of this machine which was on display at the Liberty Belle Saloon in Reno, Nevada (this business, owned by Charles Fey's grandsons Marshall Fey and Frank Fey, closed down in 2006) is shown at right.

Recently, a cigarette lighter was made with a design that resembled Charles Fey's original Liberty Bell slot machine. I'm not aware of any other toy or replica slot machine made to look like it. Some years back, there was a line of pencil sharpeners made to resemble various objects, one of which being a slot machine. One version of the slot machine looked like a Watling Rol-a-Top (the Die-cast 9619, or Play-Me 980), but the other one (the Die-Cast 9615, so they were indeed by the same company), although it looked like an early slot machine, didn't resemble any real slot machine I knew of. Recently, I unravelled that mystery; it was made to look like the "Roaring Twenties Silver Eagle Slot Machine", a 1977 toy by Poynter. The small brass pencil sharpener bore the words "Golden Eagle" below the eagle on the case.

The most commonly quoted date for the invention of the three-reel slot machine with automatic payout by Charles Fey is 1895, as given by John Scarne; however, the 1995 book Lemons, Cherries, and Bell-Fruit Gum by Richard Bueschel, a carefully-researched history of the slot machine, sets the date at 1905 instead, giving Charles Fey only a year, instead of a decade, to profit from his invention before the San Francisco Earthquake of 1906. On the other hand, Slot Machines: A History of the First 100 Years by Charles Fey's grandson Marshall Fey gives an earlier date, although still later than Scarne's, of 1898 for its invention. The first of these machines had reel strips with pictures of playing cards on them instead of just suit symbols; however, changing the reel strips and payout cards on a slot machine to use a different style of symbol is sufficiently trivial that it is not unreasonable to view the original version and the more common one as the same machine.

The contents of the reels are:

(*) Star          -   -   2
(%) Horseshoe     5   5   -
(O) Spade         2   2   2
(@) Diamond       1   1   3
(A) Heart         1   1   1
(=) Bell          1   1   2

Given its payouts,

%%.   1
%%*   2
OOO   4
@@@   6
AAA   8
===  10

one can calculate how much it returns on 1000 coins:

5 * 5 * 8 = 200    *  1 = 200
5 * 5 * 2 =  50    *  2 = 100
2 * 2 * 2 =   8    *  4 =  32
1 * 1 * 3 =   3    *  6 =  18
1 * 1 * 1 =   1    *  8 =   8
1 * 1 * 2 =   2    * 10 =  20

A return of 37.8% gives a profit of 62.2%, which is quite high.

However, Marshall Fey's book, Slot Machines: The First 100 Years, gives its payout as double that, which is more reasonable. The payout card gave the payout as a number of drinks; apparently, if the machine took nickels, a drink cost ten cents, so each drink was actually equivalent to two coins.

Note that these symbols, although different from those used on later machines, can be seen as corresponding to them: horseshoes become cherries, the star becomes the lemon, spades become oranges, the diamond becomes the plum, hearts become bells, and the bell becomes the bar.

This is made more obvious by the second set of payouts, in which all the prizes are (or, rather, appear to be, given that a drink is worth two coins, as noted above) doubled, and the feature of allowing a bell on the last reel to complete most three-in-a-row combinations is added. These are the payouts of the first slot machine made by Mills, the Mills Liberty Bell slot machine.

The third set of payouts shown is that of the Mills Novelty Company's Operator's Bell slot machine, pictured at left, which introduced the conventional fruit symbols which are now so strongly associated with slot machines, as they became popular and were quickly copied by the other companies that began to copy Charles Fey's invention. Except for the symbols used being changed, that set of payouts is the same as that offered on their earlier Liberty Bell, which used the same symbols as the original machine by Charles Fey.

Another early Mills slot machine, but still one of the early models, modified so that it wasn't raised above the counter by feet on the bottom, so as to make room for a larger coin box is pictured at right. Although these machines was quite different in appearance from Charles Fey's original machine, for a while not only later Mills machines, but other early machines from Caille, Watling, Jennings and Pace (and others, such as Cooper Manufacturing and Silver King) were much the same as this machine and the first Mills three-reel slot machine from 1907 in general appearance.

In 1931, Benton Harbor Novelty brought out their Imp slot machine, not to be confused with the Imp trade stimulator by Groetchen from 1940 (although one source shows the same device as made by Engel Manufacturing in 1927). This slot machine had the same shape as the early slot machines, but was of a somewhat smaller size, being about 16 and a half inches tall. The Pace Bantam, another smaller machine with a standard shape, was not quite as small, at 19 and a half inches in height. It is sometimes referred to as being "three-quarter size"; a full-sized slot machine of the kind to which it might be compared could be 25 inches tall. Thus, these machines serve as testimony to the enduring popularity of what came to be seen as the basic slot machine shape.

1931 was also the year when Mills brought out the Mills Silent slot machine, which decisively ended the dominance of this style and shape for slot machines. The Mills War Eagle, one of the initial styles in which the Mills Silent was introduced, is illustrated at left.

The two other initial styles of the Mills Silent are illustrated below: the Mills Front O.K. Vendor, for territories where it was necessary to incorporate a vending attachment with the slot machine, and the "Lion Front" which retained the older style of case, but with the newer silent mechanism.

The Caille Superior arrived earlier, dating back to at least 1926, and as well, by 1929, Watling slot machines were already being made in the "Blue Seal" style, but the Mills Silent was a more radical departure from earlier slot machines in styling. The Caille Superior and the Watling Blue Seal, respectively, from left to right, are pictured below:

A gum vending version of Mills' earlier Liberty Bell machine which was introduced in 1910, the same year that the Operator's Bell came out, also used the modern fruit symbols, except that while the letters of the word "GUM" appeared over the symbols in most cases, in the case of the plum, the letters were made out of many tiny plums, in the case of the lemon, three lemons on a branch were intertwined with the letter - and the cherries were replaced by a leaf, representing the spearmint flavor of gum.

Instead of a black, or dark green, bar with the famous "Bell-Fruit Gum" name on it, on that first machine with the modern fruit symbols, the bar was blue, and advertised Liberty Bell Gum Fruit.

Before the Operator's Bell, first the Liberty Bell used the same symbols as Charles Fey's original slot machine, and then an alternate version of the reel strips were made, with the substitutions of patriotic symbols shown below, and also a baseball-themed machine, pictured at right, was made quite a bit later, in 1930:

Fey            Mills (Operator's)       Mills (Liberty)      Mills (Baseball)

Horseshoe      Cherries                 Wreath               Batter
Spade          Orange                   Seal                 Baseball
Diamond        Plum                     Flag                 Glove
Heart          Bell                     Bell                 Catcher's Mitt
Bell           Bar                      Shield               Pennant
Star           Lemon                    Torch                Crossed Bats

The seal was a round paper seal such as would be affixed to a document in place of a wax seal.

Also numerous other variations, such as brands of cigarettes and dice were seen later. One rare early variation was an Operator's Bell machine by Cooper Manufacturing which replaced the bar by an Eagle.

These early slot machines did not have payouts as large as became common later. The payout schedule shown in the diagram above for the earliest of the slot machines with fruit symbols is:

%%.        2
%%*, %%A   4
OOO, OO=   8
@@@, @@=  12
AAA, AA=  16
===       20

without a jackpot as such. These slot machines had a different arrangement of symbols on the reels from that shown in the example on the previous page, because, with the smaller prizes, the machine could still be profitable, even if the players would win more often.

One such machine had this arrangement of symbols:

(*) Lemons       1   1   3
(%) Cherries     4   4   -
(O) Oranges      5   5   6
(@) Plums        3   3   5
(A) Bells        5   5   4
(=) Bars         2   2   2

With the payout scheme above, this machine's payout percentage can be calculated as follows:

Reel           Combinations  Value   Amount
 1    2    3                         Paid

 4 *  4 * 14 =          224  *   2 =   448
 4 *  4 *  6 =           96  *   4 =   384
 5 *  5 *  8 =          200  *   8 =  1600
 3 *  3 *  7 =           63  *  12 =   756
 5 *  5 *  6 =          150  *  16 =  2400
 2 *  2 *  2 =            8  *  20 =   160

for a percentage favoring the house of 28.15%. Although quite high, note that it is less than the 30.175% for the common design with the higher payouts given on the previous page. As well, that design offered 189 combinations paying 10 or more, while this design offers 421 combinations paying 8 or more. If players tend to put small prizes back into the machine, but quit while they are ahead after winning a larger prize, a machine with a low percentage, but which pays out most of the money in small prizes, is still very likely to take the money the player was willing to wager in the end. Of course, this depends on the setting; in some locations, players are likely to wager only one or two coins and move on, and then raw percentage is what matters.

Although this more generous set of payouts was introduced with machines that had twenty symbols on each reel, and thus 8,000 possible combinations, it was possible to use these payouts with a machine that, like Charles Fey's original Liberty Bell, had only ten symbols on each reel, and only 1,000 combinations. Thus, for example, the Jennings Little Duke, pictured at right, where the innermost wheel corresponded to the leftmost reel on a conventional slot machine, had this arrangement of symbols:

(*) Lemons       2   -   3
(%) Cherries     3   5   -
(O) Oranges      2   2   2
(@) Plums        1   1   3
(A) Bells        1   1   1
(=) Bars         1   1   1

The proportion of 1,000 coins spent on the machine that is paid out in prizes on average can be calculated as follows:

Reel           Combinations  Value   Amount
 1    2    3                         Paid

 3 *  5 * 10 =          150  *   2 =   300
 3 *  5 *  4 =           60  *   4 =   240
 2 *  2 *  3 =           12  *   8 =    96
 1 *  1 *  4 =            4  *  12 =    48
 1 *  1 *  2 =            2  *  16 =    32
 1 *  1 *  1 =            1  *  20 =    20

In addition to the Jennings "Little Duke", another slot machine using concentric disks was one version of the Pierce "Whirlwind". While Jennings made a conventional three-reel slot machine resembling the Little Duke which it called the Duchess, Pierce applied the Whirlwind name to both a machine with concentric disks and a conventional three-reel machine. Also, the A. B. T. Manufacturing Company made a countertop machine of a similar design called "Wagon Wheels".

Of course, in addition to these real-world examples of slot machines which use concentric discs instead of reels, the spherical alien slot machines in the Canto Bight casino in the movie The Last Jedi were of that general type as well.

An ingenious design allowing the use of three disks of equal size is used by a European maker of slot machines, for the "Rotamint" machine. Of the three symbols in a row, the leftmost one is provided by a disk the center of which is to the left of the window, and the rightmost is provided by a disk the center of which is to the right of the window. So the symbols are printed on the left disk so that if you go "upwards" you move counter-clockwise, and on the right disk, moving along the symbols in the direction of their tops is clockwise.

A gap is left between those two disks, where the middle symbol can be displayed. The third disk that supplies that symbol has its center directly below the center of the window, and the symbols are printed on it with their tops radiating outwards, just as on the three disks of the Little Duke.

The diagram below will perhaps make the principle more obvious than an explanation in words:

Most Rotamint machines had numbers on the wheels, sometimes with a limited mixture of other symbols, but at least one model, the 6 Fruit 6, used the Mills fruit symbols. Its payouts are shown to the right, and, as can be seen, they did differ somewhat in philosophy from those of American slot machines with those symbols.

The Rotamint machine turns out to have been anticipated, I recently learned, by the "4 Leaf Clover" from the Pierce Tool and Manufacturing Company in 1934. However, it did not fully anticipate all the features of the Rotamint.

It had four disks, three with fruit symbols, and one which indicated if a regular pay or a double or triple pay would be given in the event of a win. (It was a trade stimulator, so, without automatic payout, this did not complicate the mechanism.)

The three wheels with fruit symbols were identical, instead of having the symbols oriented on each wheel so that the symbols in the payout area would be upright, and the payout area was in the shape of a diamond, so the opportunity of having three symbols in a line was not used.

Although the payout structure was like that of an older slot machine, with two cherries and cherry-cherry-lemon and cherry-cherry-bell being winning combinations, as it used three identical wheels, the cherry was possible in the third position, and the lemon in any position.

Each wheel had:

Lemon      4
Cherry     6
Orange     3
Plum       3
Bell       2
Bar        2

Unlike the Rotamint, there were twenty symbols on each wheel rather than just ten.

Of course, there are many other possible ways to design an unconventional slot machine. One that I recently learned about was used by the British company Brecknell, Dolman, and Rodgers, or BDR: the three symbols displayed vertically, and were read from the top down. They were on reels, rather than discs, and the axis of the reels was directly vertical, but the reels were shaped like part of a cone with the upper reels smaller, thus allowing for a somewhat more compact machine.

Also, some European slot machines, while using conventional reels and conventional fruit symbols, had payout schedules that were very different from those of most slot machines. For example, at one time in Britain, slot machines were legal, but under some severe restrictions: to make them a relatively harmless form of gambling, the maximum payout was limited to 12 coins.

That led to one machine, the Kraft Sterling, based on a Beromat mechanism from the German firm of Gunther Wulff, as well as a few other similar ones, compensating by offering a large number of winning combinations:

Bell or     Bell or
Plum or     Orange or     (Any)          2
Cherries    Cherries

Bell or     Orange        Bell or
Plum                      Plum or        4

Plum        Bell          Bell or
                          Plum or        8

Bell        Bell          Bell or
                          Plum or       12

which was very much unlike the payout schedule of a conventional slot machine. The reels had five different symbols, one of them being a Pear, which was not part of any winning combination, the other four being Cherries, Orange, Plum, and Bell. It's possible the unusual payout schedule is partly due to a simplified payout mechanism, something like the case of the Mills QT.

A BDR slot machine - of the type where the three reels are arranged in a vertical truncated cone - despite having higher payouts and a conventional jackpot, also had a payout arrangement with a somewhat similar philosophy:

Plum           (Any)          (Any)         3

Cherries or    Orange         (Any)         3 
Union Jack

Cherries or    Cherries or    Lemon         6
Union Jack     Union Jack

Cherries or    Cherries or    Plum          9
Union Jack     Union Jack

Cherries       Cherries or    Union Jack   12
               Union Jack

Union Jack     Cherries       Union Jack   15

Union Jack     Union Jack     Union Jack   Jackpot + 18

On the first reel, the machine had banannas, oranges, strawberries, and lemons, not only the symbols which could participate in a payout, and no doubt that was true of the other reels as well.

Instead of taking a penny, another machine we will consider, the Bell-Fruit Three Penny Supreme took a three-penny coin (a coin the size of a nickel, but much thicker, made of nickel brass; like a nickel (well, at least an older, Canadian one), it was polygonal instead or round. Three sevens paid out a "Gold Award Token", so apparently merchandise prizes in excess of twelve pence were permitted.

Its payout schedule, shown at right, was like that of a conventional slot machine, but with payouts from the first two wheels as well. Despite that, with the very low payouts allowed, it's hard to see how it could have avoided a very low payout percentage.

The sevens were overprinted on other symbols; but, surprisingly, the machine also had a strawberry symbol present on the third reel.

At one point, I thought that perhaps it replaced a lemon, with the machine using elements of the mechanism of a Mills 21 Bell slot machine repurposed; the use of a non-paying symbol to limit the payout percentage seems hardly necessary under these circumstances.

However, the story is more complicated than that, as the machine did also have lemons on the third reel. While the 7 was overprinted on a bar on the third reel, as with the 21 Bell, it was overprinted on a melon on the first reel, and on a bell on the second reel, so apparently it did have a reel strip design of its own.

Today, when we think of a slot machine, one of the first things that comes to mind is the chance of winning the jackpot, and thus, a jackpot appears to be a necessary and essential feature of a slot machine, but the designs we have examined up to this point have all lacked one.

Marshall Fey (a grandson of Charles Fey, the inventor of the slot machine), in his book "Slot Machines: A History of the First 100 Years", notes that Watling produced a slot machine in 1920 that included a jackpot, but it did not become popular. He also notes that subsequently, Rock-Ola modified existing slot machines to add a jackpot starting in 1927, but it was after Caille, in 1928, introduced, and enjoyed success with, the Caille Superior Jackpot Bell that the other slot machine makers quickly followed suit. This led to the present situation, where a jackpot was something people expected from a slot machine, coming about in short order.

The fourth set of payouts, increasing the prizes for three oranges, three plums, and three bells from 8, 12, and 16 to 10, 14, and 18 respectively, which soon became the standard of the industry, was introduced with the Mills Mystery slot machine from 1937, also known as the "Castle Front", illustrated at left, which also offered some extra "mystery" payouts not explicitly listed. (Some of the lemons on the first reel were treated as cherries, those were the mystery payouts.)

The prize for two cherries was also increased from 2 to 3, and the prize for two cherries and either a lemon or a bell was increased from 4 to 5. These increased payouts also became common on competing slot machines, but there the older 2/4 payouts continued to coexist with the more generous 3/5 payouts; this was particularly true after single-cherry pay was introduced.

This set of payouts is that discussed in the example on the page on probability. This was the standard set of payouts for a slot machine; for many years prior to World War II, nearly every slot machine made used those payouts. Three bars paid 20 coins plus the jackpot, depicted here as 100 coins.

Since the amount of the jackpot may vary due to how long it has been since the most recent previous jackpot has been won, of course this may be an oversimplification.

Speaking of mysteries, what must people have thought in November, 1948, when the new version of Watling's venerable Rol-a-Top slot machine, pictured at right, appeared?

Although it would only be on July 25, 1950 that the Korean War would start, with an invasion of South Korea by North Korea, and it was only in September, 1949 that President Truman announced to the American people that the Soviet Union had the atomic bomb, conserving metal for defense purposes, with World War II having only recently concluded, and with some aspects of the Cold War dating from 1947 as the United States reacted to Russia's attempts to obtain political control of Eastern Europe, by using pieces from old Mills slot machine cases might not have been seen as either unreasonable or mysterious.

However, that only explains the left side of the front, and not the right side, so apparently this is a bizarre case of unabashed copying.

From the manual for the Jennings Chief slot machine, a typical way in which the jackpot is filled on a slot machine is this:

Coins paid into the machine are first placed in the machine's coin tube if that is not full. The coin tube is the source from which coins are counted out to pay awards of a fixed amount in the payout schedule, so ensuring that there are enough coins in it has the highest priority.

If a coin passes the coin tube, next it goes into the jackpot if that is not full.

If the jackpot is full, it goes into the reserve. After a jackpot is paid, the contents of the reserve go into the jackpot so that the machine is not left with an empty jackpot, which would not encourage players.

Finally, when both the jackpot and reserve are full, coins go into the machine's cash drawer, where they serve as the operator's income. Although it's not all profit yet, since occasionally it will be worthwhile refilling the reserve or the jackpot by hand.

In this scenario, where no time is lost in keeping the jackpot full, calculating the percentage as if jackpots always paid the full capacity of the machine's jackpot chamber is indeed a reasonable approximation.

For convenience, I will repeat the calculation for this example slot machine from the first page of this section here:

The arrangment of symbols on the reels for those payouts which has been cited in several books on gambling was:

(*) Lemons       3   -   4
(%) Cherries     7   7   -
(O) Oranges      3   6   7
(@) Plums        5   1   5
(A) Bells        1   3   3
(=) Bars         1   3   1

designed for a machine with the schedule of payouts as depicted in the diagram above:

%%.        3
%%*, %%A   5
OOO, OO=  10
@@@, @@=  14
AAA, AA=  18
===      120

And the calculation of its return proceeds as follows:

Reel           Combinations  Value   Amount
 1    2    3                         Paid

 7 *  7 * 13 =          637  *   3 =  1911
 7 *  7 *  7 =          343  *   5 =  1715
 3 *  6 *  8 =          144  *  10 =  1440
 5 *  1 *  6 =           30  *  14 =   420
 1 *  3 *  4 =           12  *  18 =   216
 1 *  3 *  1 =            3  * 120 =   360

for a profit of 24.225%.

I have now found more information about where this particular example of a slot machine came from. This set of symbols on the reels is given in the book "The Facts of Slots" along with more detail: it belonged to a slot machine confiscated by the police that was then given to a statistics professor for legitimate use in education. This professor, Dr. Philip G. Fox, is shown in a newspaper photograph available online with a slide rule in his hand examining a Mills Extraordinary, like the one pictured at right; of course, I can not be certain if that is the machine mentioned in connection with this particular distribution of symbols on the reels.

This professor and his slot machine were also referenced in the April 14, 1945 issue of Billboard magazine.

An actual machine, the Pace Comet, pictured at right, in at least one version, had a very similar arrangement.

The Pace Comet in the picture, called the Fancy Front, is from 1934, around the same time that Watling introduced its new Rol-a-Tor slot machine. The Watling Rol-a-Top, as it was soon to be named, had an elaborate rotary escalator that showed the last eight coins played (perhaps less for some denominations); Pace slot machines already, at least since 1926, had a smaller rotary escalator which showed the last four coins played.

The arrangement of symbols on the reels shown below is given on this page.

(*) Lemons       3   -   5
(%) Cherries     7   7   -
(O) Oranges      3   6   6
(@) Plums        5   2   5
(A) Bells        1   3   3
(=) Bars         1   2   1

Another version of the Pace Comet, this time the All-Star Comet, from 1937, and contemporary with the Pace Kitty, to be discussed below, is shown at right.

With this arrangement, and the following similar payout schedule:

%%.        3
%%*, %%A   5
OOO, OO=  10
@@@, @@=  14
AAA, AA=  18
===      150

we can calculate its percentage:

Reel           Combinations  Value   Amount
 1    2    3                         Paid

 7 *  7 * 12 =          588  *   3 =  1764
 7 *  7 *  8 =          392  *   5 =  1960
 3 *  6 *  7 =          126  *  10 =  1260
 5 *  2 *  6 =           60  *  14 =   840
 1 *  3 *  4 =           12  *  18 =   216
 1 *  2 *  1 =            2  * 150 =   300

for a profit of 20.75%.

Even this more generous set of payouts does not strictly require having 20 positions on each reel, as can be illustrated through the example of the Groetchen Columbia (pictured at right):

(*) Lemons       3   -   3
(%) Cherries     2   5   -
(O) Oranges      2   2   2
(@) Plums        1   1   3
(A) Bells        1   1   1
(=) Bars         1   1   1

Note that the arrangement of symbols on it is very similar to that on the Little Duke, altered only by replacing one cherry on the first wheel by a lemon! The fraction of the money spent on the machine that is returned in prizes can be calculated as follows:

Reel           Combinations  Value   Amount
 1    2    3                         Paid

 2 *  5 * 10 =          100  *   3 =   300
 2 *  5 *  4 =           40  *   5 =   200
 2 *  2 *  3 =           12  *  10 =   120
 1 *  1 *  4 =            4  *  14 =    56
 1 *  1 *  2 =            2  *  16 =    32
 1 *  1 *  1 =            1  *  20 =   120

Thus, it would still be profitable even if the jackpot averaged 100 coins; in practice, jackpots were no doubt smaller.

The Mills SP set of reel strips is also similar to this:

(*) Lemons       3   -   5
(%) Cherries     7   7   -
(O) Oranges      3   6   6
(@) Plums        5   2   5
(A) Bells        1   3   3
(=) Bars         1   2   1

in fact, it is identical to what was found on the version of the Pace Comet cited above.

The Mills Mystery slot machine, however, also had additional mystery payouts. The mystery payout was that two of the three lemons on the first reel actually behaved as if they were cherries. So, if it used the Mills SP reel strips, we can calculate what it would actually have paid:

Reel           Combinations  Value   Amount
 1    2    3                         Paid

 7 *  7 * 12 =          588  *   3 =  1764
 7 *  7 *  8 =          392  *   5 =  1960
 3 *  6 *  7 =          126  *  10 =  1260
 5 *  2 *  6 =           60  *  14 =   840
 1 *  3 *  4 =           12  *  18 =   216
 1 *  2 *  1 =            2  * 120 =   240
Mystery pays:
 2 *  7 * 12 =          168  *   3 =   504
 2 *  7 *  8 =          112  *   5 =   560

So instead of paying out only 6280 coins out of every 8000, it also paid out an additional 1064 coins, for a total of 7344 coins and a payout percentage of 91.8%, leading to a house advantage of 8.2%.

The fifth set of payouts is for the Mills Golden Bell slot machine, sometimes referred to as the Mills Roman Head, pictured at left. This is one of the first machines to introduce an extra symbol which stood for an additional jackpot. The machine dispensed a brass token which was then redeemed by the operator for this additional payout, called the Gold Award, which was therefore of a fixed size, unlike normal jackpots.

As for the normal jackpot, I've stumbled upon an early advertisement for the Mills Golden Bell slot machine which gives detailed information about its regular jackpots. The capacity of the jackpot compartment depended on the size of the coin the machine was handling, as is not unreasonable. Thus, it could handle 96 nickels ($4.80), 104 dimes ($10.40), or 84 quarters ($21.00).

This innovation was soom adopted by some other slot machine makers, particularly by Watling, which, from 1934 through 1949 manufactured the Watling Rol-a-Top slot machine. This is considered to be one of the most beautiful slot machines ever made, thanks to its elaborate circular coin escalator (the part that shows the last few coins played to help prevent the use of slugs).

Actually, when it first came out at the end of 1934, it was called the Rol-a-Tor, but within a few months its name was changed. Although apparently there is no documentary evidence in the matter, it is generally assumed that this was because the original name was too reminiscent of the Norge Rollator refrigerator, popular at the time. Thus, at the right, the Watling Rol-a-Top slot machine is shown in all its glory, along with a version of the original Rol-a-Tor where a vendor front covers part of the case. In addition, to the left of the image, the Watling Treasury, very similar to the Rol-a-Top, except for the lack of the large coin escalator, is shown for comparison.

While the Rol-a-Tor, with minor variations, continued to be sold until 1949, in late 1938 Watling appears to have discontinued production of the Treasury; one advertisement stated that they had only 45 machines remaining in stock.

In addition to the two columns of mints on either side of the jackpot compartment, it may be noted that the eagle in the case design is replaced by something above the jackpot compartment. And that is the window through which the machine's supply of Gold Award tokens is displayed, so the vendor model shown here is one with the Gold Award feature under discussion.

The Mills Golden Bell has a similar window; it can be seen to the right of the gold coin motif in the center of the case, starting from its vertical center, and extending down to the head of the female figure on the right at the bottom of the case.

Or, to make it simple, rather than requiring the reader to follow lengthy verbal descriptions in each case, here are the Mills Golden Bell and the Watling Rol-a-Tor again, but this time with the Gold Award Token windows highlighted in yellow in each one:

And following those two images is one of the Watling Gold Award slot machine from 1934, as this did predate the Rol-a-Top. No highlighting was performed on that illustration, but the window displaying the Gold Award tokens is in the same position as on the Rol-a-Tor vendor. (The small size of the image is due to the source image being of poorer quality, not because it is a smaller machine.)

Jennings also made slot machines which had a symbol representing the token given out to be redeemed for an additional fixed large prize; while Watling also used the term "Gold Award", and Pace did as well, Jennings term for a prize tokens was "Prosperity Token", which, of course, would be an appropriate name to use during the Great Depression.

The illustration below

shows the payouts for the Mills, Watling, Jennings, Pace and Superior Confection machines with gold award tokens in order, from left to right. The Mills and Pace machines had the same payout scheme, with three bars awarding both a normal jackpot and the constant prize of 20 coins together; the Jennings machine just awarded a jackpot, and the Watling machine apparently did not also award a jackpot on three bars.

The fifth payout schedule is from a machine by the Superior Confection Company. Here, the Gold Award token bears a Liberty Bell symbol on it. Although the payout schedule follows that of the Watling machine in the illustration, the payout card shows the style of Gold Award token used in a Mills machine, which suggests that some Mills Gold Award machines also used that payout schedule. Many of this company's machines were made from parts from both Watling and Mills slot machines, and on some of them, it is printed on the case that they were built for the company by Charles H. Fey and Company. Machines by the Superior Confection Company are quite rare, and are highly prized by slot machine collectors. One unique machine of theirs had a conventional slot machine mechanism oriented so as to use carousel-like reels with colored silhouettes of race horses.

The machine did have a jackpot window, so it is possible that three bars did award a jackpot in addition to the basic prize of 20 coins.

On the right is shown the payout schedule for another machine by the Superior Confection Company from 1934 or 1935 that had a "Gold Award" symbol on its reels, the Special Mystery, but it's not clear if an actual token was involved in this machine's payouts, since all the combinations of a bar and a gold award symbol are stated to pay the jackpot, and the machine has a jackpot window.

However, the machine pictured at left has both a pair of jackpot windows, and a token window immediately above the jackpot windows, in addition to an escalator above the reels (which makes it completely unambiguous that there is also a token window present).

Since the Gold Award token had an image of a Liberty Bell on it, I suppose that it was decided to have the Gold Award symbol, as well as the lemon and the bell, complete the 5-coin combination with two cherries to prevent confusion and arguments.

The only difference between the payout diagrams for Mills and Pace are that the former represents a token with an image of an amphora, and the latter a token with an image of a wreath. The actual Pace tokens, though, do not seem to have resembled the symbol used, unlike the case for the other three manufacturers.

These are just the makers of slot machines that used a symbol which was an image of the prize token on the reels of their slot machines; the use of a token for a prize payout is common, being one of the standard options in slot machine design. Later on, when guaranteed jackpots were paid by attendants in casinos, with tokens not being involved, the Gold Award symbol of a coin was replaced by other jackpot symbols such as the 7 or the melon, but even so, it can be considered to be their ancestor.

Watling's original design for its gold award token was somewhat unfortunate: on one side, it included a horseshoe... and another good luck symbol, this being one used by Native Americans, and which also had positive religious meaning in India. Unfortunately, it also was used as the symbol of a certain German political movement which fell into deserved disrepute.

Arbitrarily, the jackpot for three bars is depicted as 50 coins; winning the Gold Award provided either a larger jackpot, or a token redeemable for 100 times the amount wagered on different machines made by Mills with the new symbol.

The version of this machine with a low percentage had the following arrangement of symbols:

(*) Lemons       -   -   3
(%) Cherries     8   7   -
(O) Oranges      3   4   5
(@) Plums        2   4   5
(A) Bells        2   3   4
(=) Bars         3   1   2
(o) Coins        2   1   1

an alternate version replaced two of the cherries, and one bar, on the first reel with lemons.

Given the payouts of

%%.         3
%%*, %%A    5
OOO, OO=   10
@@@, @@=   14
AAA, AA=   18
===        70
ooo       120

we can calculate how many coins this paid out for each 8000 coins put in, on average:

Reel           Combinations  Value   Amount
 1    2    3                         Paid

 8 *  7 * 13 =          728  *   3 =  2184
 8 *  7 *  7 =          392  *   5 =  1960
 3 *  4 *  7 =           84  *  10 =   840
 2 *  4 *  7 =           56  *  14 =   784
 2 *  3 *  6 =           36  *  20 =   720
 3 *  1 *  2 =            6  *  70 =   420
 2 *  1 *  1 =            2  * 120 =   240

for a profit of 10.65%.

Although companies like Bally and Keeny were the best-known makers of console slot machines, while Mills was mostly famous for its regular slot machines, Mills also made console machines. One machine of theirs, illustrated at left, which did something interesting with its payout schedule was their "Four Bells" console.

One set of three reels showed the combination of symbols which might indicate a prize will be awarded, as on a conventional slot machine.

But the player had a choice of four different coin slots in which to wager, each one giving a different playout schedule.

Essentially, starting from a conventional slot machine payout schedule, one could choose to double the payout on a combination of oranges by using the first slot, or to double the payout on a combination of plums by using the second slot, or to double the payout on a combination of bells by using the third slot... or to triple the size of the jackpot payout by using the fourth slot.

Note that the payout for two cherries is 2, and the payout for two cherries and a lemon or a bell is 4, as on older slot machines, but the basic payouts for the orange, plum, and bell are 10, 14, and 18 respectively, the larger payouts introduced with the Mills Mystery.

So, let us start with the Mills SP set of reel strips, designed for use with the smallest payouts being 3 and 5, and work out what it would pay with them reduced to 2 and 4, and with 60 coins being paid for three bars.

Reel           Combinations  Value   Amount
 1    2    3                         Paid

 7 *  7 * 12 =          588  *   2 =  1176
 7 *  7 *  8 =          392  *   4 =  1568
 3 *  6 *  7 =          126  *  10 =  1260
 5 *  2 *  6 =           60  *  14 =   840
 1 *  3 *  4 =           12  *  18 =   216
 1 *  2 *  1 =            2  *  60 =   120

If the amount paid for three oranges, or two oranges and a bar, was doubled, then add 1260 coins to the amount paid, for 7040 coins out of 8000.

If the amount paid for three plums, or two plums and a bar, was doubled, then add 840 coins to the amount paid, for 6620 coins out of 8000.

If the amount paid for three bells, or two bells and a bar, was doubled, then add 216 coins to the amount paid, for 5996 coins out of 8000.

And, finally, if the amount paid for three bars was tripled, add 240 coins to the amount paid, for 6020 coins out of 8000.

Thus, the SP set of reel strips would be sufficient to lead to profitability for the machine, although it would have been better to use a customized set of reel strips that led to the house advantage for the four options presented to the player being more equal.

Some older slot machines, to lower costs, used simplified arrangements.

Thus, the Mills QT (and the Mills Junior Silent, on which it was based) paid out only when two cherries appeared on the first two reels, with the prize determined by the symbol on the final reel, except for the jackpot which was still paid on three bars.

That machine's payout schedule is pictured at right, and using the symbols used above in the text, it was:

%%=  2
%%O  2
%%@  2
%%*  4
%%A  8
%%% 12
=== Jackpot

which could also be viewed as

%%.  2
%%*  4
%%A  8
%%% 12
=== Jackpot

which suggests that the design is partly based on abbreviating the payout mechanism of a conventional slot machine, rather than on a simpler design of a completely different nature. Differentiating between the lemon, bell, and cherry on the third reel, though, still requires something additional.

The machine itself is pictured at right.

Small-sized machines like the Mills QT stopped being manufactured abruptly just prior to 1951. In the United States, up to that time, slot machines were commonly found in ordinary retail establishments, frequently even when this was in violation of local ordinances or even State laws. Very early in 1951, however, the Johnson Act took effect, a Federal law which not only imposed severe penalties for illegal slot machine operation, but also made their manufacturers responsible to only ship machines to those who could legally use them.

This meant that State laws restricting slot machines became effective; in addition to Nevada, only three other states at the time permitted slot machines to be operated for gambling, two only in private clubs, and the third under very limited conditions as well.

The Single-Cherry Era

Up to this point, slot machines nearly all paid one prize for two cherries, and a larger prize for two cherries and either a bell or a lemon. As we've seen, this arrangement predates the introduction of fruit symbols on Mills machines, going right back to Charles Fey's original slot machine on which two horseshoes returned the coin played, and two horseshoes and a star paid two coins. But it was the first Mills slot machine that added a payout for two horsehoes and a heart, the symbol taking the role that the bell would later occupy.

The fourth column of the illustration below shows the payout schedule that post-war Jennings machines eventually came to offer, both with an award for one cherry on the first reel, and then two other awards, increasing in size, for two cherries and then all three.

Before slot machines started to give prizes for a cherry on the first reel only, and before they started to have cherries on the third reel, allowing the lemon to be done away with, Mills emphasized the importance of the cherry in a different way with the machine known as the "Bursting Cherry" from 1938.

This machine, pictured at left, had a conventional pre-war payout schedule, except that the award for two cherries and either a bell or lemon was ten coins instead of five, as shown in its payout schedule, in the first column of the illustration above.

Apparently, the 1939 Mills Chrome Bell slot machine, pictured at right, (often referred to as the Diamond Front) was the first machine to offer payment on one cherry on the first reel only. I have been able, so far, to directly confirm the machine's existence in May 1939 and single-cherry pay being offered on it at least by October 1940. Its payouts are illustrated in the second column of the illustration above.

After the Second World War, some companies sold both old Mills Diamond Front Chrome Bell machines and other Mills Silent Bell machines with cases made to resemble the case of that machine, and those other machines were modified to also have a single-cherry payout.

Also, after the war, the first new Jennings machine was the Bronze Chief, followed by the Super Deluxe Club Chief and then the Standard Chief. These machines were available with several options for their payouts and reel strips, and single-cherry payout was included. This form of payout for those machines is illustrated in the third column of the illustration above. In fact, I have since found an advertisement from September 1939 in which it is noted that the pre-war Super Chief from Jennings is also offered with one-cherry pay as an option. The wording of that advertisement suggests to me that they were offering an alternate payment option because a competing machine - specifically, the Mills Chrome Bell - had already made it popular, so the fact that it pre-dates the earliest mention I could find of the Mills machine's single-cherry pay does not make me doubt that Mills originated it on that machine. Also, while most (but not all) surviving Mills Chrome Bells owned by collectors have single-cherry pay, all the Jennings Super Chiefs that I could find images of had the older double-cherry payouts.

At first, the two payouts, on one cherry and two cherries, simply replaced the former two payouts, on two cherries, and on two cherries with a lemon or bell.

The War on Slugs

and other matters connected with coins and their denominations

Only a few of the advertisements for the Jennings Super Chief, pictured at left, mentioned the option of single-cherry pay for it. Its primary innovation, however, was prominently featured in all the advertisements for it: an improved slug rejector. They seemed to be so confident in that improvement that the machine did not even have a coin escalator; despite that, Jennings' next machine, the Jennings Master Chief, not only also included a coin escalator, but included a large and bulky coin rejector supplied to Jennings by another company, so apparently their improvements in the Super Chief still fell short of perfection.

In the illustration of the Super Chief on the left, there is something below the payout card which looked as though it might be a round window through which at least the last coin played could be seen, if not a full multiple-coin escalator. However, from photographs of the machine, it appears to be something else; what appears to be a coin showing through a round window is instead a metal projection which would have obstructed the view of a coin behind it. (One possibility is that the merchant has to push it down and see the last coin played behind it, as above the projection there still appears to be a dark area which may be a window, before a larger payout is made by the machine, which would mean that a key or at least a concealed control somewhere would also have to be used.)

Apparently, slugs were a serious problem im late 1939, as Pace also advertised their Rocket Bell as "slugproof", and improved slug rejection was also advertised for jukeboxes by Rock-Ola. Earlier in 1939, Pace offered a version of their De Luxe Comet Bell which also had a large additional slug rejector added, and that too is illustrated below. Like the Jennings Super Chief, it appears to have omitted the coin escalator, and the escalator came back on its successor, the Rocket Slugproof Bell.

Below are illustrated the Jennings Silver Chief from 1937 (but which was still being advertised later on, along with the Jennings Super Chief), the Jennings Master Chief which shortly followed the Jennings Super Chief, the Jennings Silver Moon Chief from 1941, and the Jennings Silver Moon Master Chief:

And here are illustrated the Slug Rejector version of the Pace de luxe Comet Bell as well as the Pace Rocket Slugproof Bell:

For the Jennings Master Chief, the company turned to a third-party company, National Rejectors Incorporated of St. Louis, Missouri, to supply the "National slug rejector" which, as can be seen, is somewhat bulky and obtrusive in appearance on this machine. While the regular Silver Moon Chief did not include this slug rejector, there was also the Silver Moon Master Chief which did, and both are shown above.

I have learned why slugs were such a serious problem in 1939. It was not until December, 1940 that a court decision found that those who made and used slugs could be convicted under Federal counterfeiting laws. Later, in 1944, Federal legislation specifically dealing with slugs made to defraud vending machines, distinguishing that from counterfeiting, was enacted. Before the 1940 court decision, there were novelty companies openly making slugs, although with denials that they were intended for fraudulent use.

By 1946, Pace was advertising that its machines were available with a choice of payouts, including options where one cherry on the first reel paid 3 coins, and by 1947, Watling was doing so as well, with the option of paying out either 2 coins or 3 coins for a cherry on the first reel. As we will see shortly, both of these companies also moved quickly to offer the option of payouts similar to those of Buckley's Criss-Cross shortly after Jennings came out with their similar Tic-Tac-Toe.

The Pace Deluxe Chrome Bell, pictured at left, and its predecessor, the very similar Pace Deluxe Cherry Bell, pictured at right, illustrate something else about the escalator on slot machines. The Pace Deluxe Chrome Bell pictured here has a circular coin escalator, while the Pace Deluxe Cherry Bell pictured here has a window through which the last coin played can be seen. This is not, however, because the circular coin escalator was introduced with the Chrome Bell; both models were the same in respect of coin escalators or the lack thereof.

Instead, because the 50 cent piece is a physically larger coin than a quarter, it was no longer possible to have the circular coin escalator for that denomination. Both machines were also made in a version that handled silver dollars; there, automatic jackpot payout was also dropped, and instead a gold award token was used.

The earliest slot machines were only made to handle nickels. Some slot machines were available to take pennies; this depended on the date, and the type of machine. From 1930 to 1950, most full-sized slot machines were made in versions that took nickels, dimes, and quarters. The Watling Rol-a-Top was initially made in a model that took pennies as well. Also, on rare occasions during the history of that machine, Watling advertised a 50 cent version of the Rol-a-Top, sometimes noted as available at a higher price. Thus, a 50-cent Rol-a-Top is a rare collector's item.

The same physical considerations that Pace faced also confronted Watling: when it decided to make a machine that took silver dollars for casino use, that machine was a version of its Treasury rather than its Rol-a-Top.

The Jennings Chief and some of its successors during the 1930s were advertised as being available for all denominations from the penny to the silver dollar, and in the early post-war era, Pace advertised its machines as being available for all denominations from the nickel to the silver dollar. Caille, and also Mills, tended not to mention available denominations in their advertisements. However, in the early post-war era, some Mills distributors noted that the pre-war Mills Horsehead Bonus slot machine, to be discussed below, had just become available for the 50 cent coin.

Of course, the first slot machine to accept silver dollars was the Silver Dollar from 1929, a revamp made by Charles Fey; and, of course, it was unfortunately timed, as the prosperity of the Roaring Twenties was about to come to an end with the stock market crash in October of that year.

Some console machines had multiple coin slots, so that one could play either a nickel, a dime, or a quarter, receiving one's prize in the coin that one played. Bally made a countertop slot machine which could take either a nickel or a quarter, the Bally Double Bell, from 1939, shown at left, and this was followed by the Keeney Pyramid in 1950, shown at right, which was a console, but a console in the sense of a floor model of a conventional slot machine like the Caille Commander, rather than a wide machine like the Mills 3 Bells.

The Jennings Triplex Chief from 1938, pictured at left, tried another approach: it was a nickel slot machine, but playing a dime would let you play twice, and playing a quarter would let you play five times. This turned out to be a mistake; if enough people played dimes and quarters, the machine was likely to run out of nickels.

That didn't stop Watling from making the same mistake: a very rare model of the Rol-a-Top could take quarters as well as nickels, and give five plays for a quarter: it was advertised in both the May 7, 1938 and the May 14, 1938 issues of Billboard. Just to show I'm not making this up, the ad is shown on the right.

Incidentally, a rotary coin escalator such as featured on the Pace Deluxe Chrome Bell shown above also appeared on a Pace Operator's Bell from late 1926, illustrated at left (the new 1927 model, this date being visible on the front), so their rotary escalator antedated that of the Watling Rol-a-Top.

Cherries on the Third Reel

Returning to the Single-Cherry Era

In addition to Jennings, Pace also made slot machines for casino use with cherries on the third reel, and this was common on Bally electromechanical slot machines. But neither Mills nor Watling appear to have ever adopted the cherry on the third reel (not counting the Mills QT, of course).

While I am not inclined to count the Mills QT as being the originator or even the inspiration of including the cherry om the third reel as a routine thing on conventional slot machines, as its payout structure was very much unlike that of a conventional slot machine, I have found that a traditional payout schedule (that is, one that does not begin to give prizes for cherries until they appear on the first two reels), but with three cherries and two cherries and a bell giving the next prize after two cherries, was a feature of a conventional slot machine from 1937, the Caille Playboy, pictured at left, which was basically a version of the 1936 Caille Cadet with different payouts.

Advertisements for this machine made it clear what the rationale was for replacing two cherries and a lemon with three cherries was: abolishing the lemon completely from the reels of the machine.

The payout schedules for three machines that introduced changes to the three-reel combinations involving cherries are pictured at right, first that of the Caille Playboy on the left of the image, and then that of Keeny Bonus Super Bell in the center of the image, and finally that of Evans' Lucky Lucre on the right of the image. I am not sure if the two payouts for cherries on the Lucky Lucre are 2 coins and 4 coins, as shown, from the images of that console I have been able to locate. Also, for both machines, I am arbitrarily using 100 coins to represent the jackpot. And unlike the Evans Lucky Lucre, this was indeed specifically done so that the machine would have no lemons anywhere on its reels.

Of course, as this slot machine dates from 1937, its name has nothing to do with the famous magazine founded by Hugh Hefner in 1953.

Before I became aware of the Caille Playboy, I learned that a console machine from 1946, the Keeney Bonus Super Bell, pictured at left, had a cherry on the third reel. It did not, however, pay out on a cherry on the first reel only, but it did have one payout for cherries on the first two reels, and a larger one for cherries on all three reels. While the Caille Playboy of 1938 anticipated it in putting cherries on the third reel, and paying out specifically on three cherries, in one respect this was closer to the more modern payout schedules with three cherries in that it did not also include the older cherry-cherry-lemon payout.

I had thought that subsequently, it became the norm to allow a bar on the third reel to substitute for a cherry on the third reel as with other symbols, but I see I was mistaken: that did not happen, not even in the electromechanical era. However, one console machine, the Evans Lucky Lucre, from 1939, which had a traditional payout structure, paying one prize for two cherries, and a larger prize for two cherries and a bell, also paid that latter larger prize for two cherries and a bar, rather than acknowledging lemons on the machine at all.

Also, I have since learned that payouts on one, two, or three cherries were offered on the Jennings Standard Chief after all, which means that Jennings is likely to have been first with the fully modern payout schedule, and they could have had it by 1946 as well, although at the moment the earliest data I have is 1947.

Ironically, while the Lucky Lucre broke with tradition by changing cherry-cherry-lemon to cherry-cherry-bar, so that its payout schedule would neither create a need for lemons, nor admit to the presence of lemons, an image of the reel bundle for that machine reveals that it did indeed have lemons on both the first and third reels.

While I hadn't seen any indication of this in contemporary advertising, after seeing one Jennings Standard Chief from 1947, claimed to have its original payout card and reel strips, with payouts on one, two, or three cherries, I have searched for images of other Jennings Standard Chief machines on the Web, and found several others with that payout arrangement.

As well, while changing from the traditional payout on two cherries or two cherries and an additional symbol to payout on one or two cherries can be done by simply bending back a payout finger in addition to changing reel strips and payout discs appropriately, changing to pay on one, two, or three cherries requires a fundamental addition to the mechanism.

Thus, Jennings definitely appears to have originated the modern payout structure with prizes on one, two, or three cherries, and it appears to have existed from the early post-war era: previously, I had despaired of narrowing it down to any time before the introduction of the Jennings Buckaroo.

The sixth arrangement shown on the first row shows the payouts of one version of Mills' Melon Bell machine. Here, the new symbol, a half-watermelon, simply replaced the bar, as can be seen from the fact that it can complete three oranges, three plums, and three bells. The original Mills Melon Bell, which came out in 1937, did not have a payout for a single cherry on the first reel; its case design was similar to the "Bursting Cherry" style of machine. Pictured at left is the original version, at right the "hi-top" version from 1948.

The window to the jackpot compartment replaced the cherry prominently featured on the case of the Bursting Cherry version of the Mills slot machine.

I have been able to verify, from some Mills advertising literature (which is better left undescribed in certain aspects) displayed in a forum post, that Mills did indeed intend the melon symbol to represent a watermelon; and looking things up on Wikipedia convinced me that at least arguably a watermelon is a true melon alongside the honeydew melon and the cantaloupe.

Some Casino Machines

Now we will look at the 21 Bell slot machine, also known as the 7-7-7 HiTop machine, which had a few positions on the three reels where a 7 was overprinted on another symbol, so that any winning combination either symbol might make would pay out.

(*) Lemons       -   -   5
(%) Cherries     3   6   -
(O) Oranges      5   4   2
(@) Plums        8   3   3
(A) Bells        1   5   8
(C) Melons       1   1   1
(=) Bars         2   1   1
(7) Sevens       1   1   1

Although it does not affect the odds, the 7 on the first reel was overprinted on a plum, the one on the second reel was overprinted on an orange, and the one on the third reel was overprinted on the bar.

Again, an alternate version of the first reel was available for a less generous payout; it had three lemons on the first reel, replacing two plums (including the one overprinted with a 7) and one of the bars.

Given the payout schedule:

%..              2
%%.              5
OOO, OO=        10
@@@, @@=        14
AAA, AA=        18
CCC, CC=, ===  120
777            200

its percentages can be calculated as follows:

Reel           Combinations  Value   Amount
 1    2    3                         Paid

 5 * 14 * 20 =         1400  *   2 =  2800
 5 *  6 * 20 =          600  *   3 =  1800
 5 *  4 *  3 =           60  *  10 =   600
 8 *  3 *  4 =           96  *  14 =  1344
 1 *  5 *  9 =           45  *  18 =   810
 1 *  1 *  2 =            2  * 120 =   240
 1 *  1 *  1 =            1  * 120 =   120
 1 *  1 *  1 =            1  * 200 =   200  

As this gives an even lower profit of 1.975%, which would seem to be impractical for any slot machine, even in a major casino, I can only conclude that I have misunderstood my sources of information. Cutting the jackpot in size from 100 coins to 50 changes the percentage to 2.95%, still extremely low.

In John Scarne's Complete Guide to Gambling, a detailed description is given of a Mills 21 Bell machine with a modified payout schedule and modified reel strips used in the Nugget casino in Reno. On this machine, there was a 7 on its own on the first reel; as in the regular Mills 21 Bell, the 7 was overprinted on an orange on the second reel, and on a bar on the third reel. But in addition, a bell and bar were overprinted on the first reel, a plum and bar along with a melon and orange in two positions were overprinted on the second reel, and a melon and orange were overprinted in two positions on the third reel. (Overprinting the melon and orange on the second and third reel makes me wonder if there is a connection to the Jennings 7-1-1 slot machine, where the Indian head, an additional jackpot symbol similar to the melon, can substitute for an orange on the last two reels.)

Recently, I came across a listing for a Jennings Standard Chief slot machine which had the number 7 overprinted on a symbol on the first reel, and the number 1 overprinted on a symbol on each of the second and third reels. Having those numbers appear, spelling out 7-11, on the pay line gave a large jackpot.

Perhaps this was Jennings' answer to Mills' 21 Bell slot machine, or it could have been a customization performed by one particular casino. But it certainly would make sense, given Jennings' use of 7 and 11 on the Jennings Sweepstake Chief, that they used 7 and 11 as a motif on other machines.

I have now found two other examples of this kind of machine. One was a Sun Chief model from 1949 instead of a Standard Chief, and it was in use at the Flamingo. The other was also a Standard Chief, but instead of having the extra 7-11 jackpot added to a conventional payout schedule, it had it added to a Tic-Tac-Toe payout schedule.

Also, in both cases, the images were more detailed, allowing me to see that in addition to the jackpot for the superimposed 7-1-1 numbers, there were additional prizes involving an additional symbol on the reels: an Indian head.

The payouts of the Mills 21 Bell, along with both variants of the payout for this machine, are shown at right.

How could such a generous payout schedule be profitable?

The one with payouts added to a Tic-Tac-Toe payout schedule looks particularly daunting, so for our example, let us choose the simpler case shown on the left of the diagram.

Let's start with an actual set of reel strips for the original payout schedule (V12-55-3), exclusive of the additions that make the machine a 7-1-1 slot machine. This set is said to have a modest payout percentage of 75%.

(*) Lemons       4   0   5
(%) Cherries     2   6   0
(O) Oranges      5   5   3
(@) Plums        5   3   3
(A) Bells        1   5   8
(=) Bars         3   1   1

For the second and third reels, the obvious step that suggests itself would be to take one orange on each reel, and change it to an Indian Head; this would leave the amount paid for three oranges unchanged.

On the first reel, there are three bars; change one of those to an Indian Head, and the amount paid for jackpots is unchanged.

And adding a single jackpot of 200 coins to a machine with a house advantage of almost 25% by overprinting those digits on one symbol on each reel would not make it unprofitable.

Thus, the reel strips might have the following contents:

(*) Lemons       4   0   5
(%) Cherries     2   6   0
(O) Oranges      5   4   2
(@) Plums        5   3   3
(A) Bells        1   5   8
(=) Bars         2   1   1
(4) Heads        1   1   1

with the payouts calculated as follows:

Reel           Combinations  Value   Amount
 1    2    3                         Paid

 2 * 14 * 20 =          560  *   3 =  1680
 2 *  6 * 20 =          240  *   5 =  1200
 5 *  5 *  4 =          100  *  11 =  1100
 5 *  3 *  4 =           60  *  13 =   780
 1 *  5 *  9 =           45  *  18 =   810
 2 *  1 *  1 =            2  * 100 =   200
 1 *  1 *  1 =            1  * 100 =   100
 1 *  1 *  1 =            1  * 200 =   200

for a payout percentage of 75.875%.

So, not only is it feasible to make a profitable arrangement, but it isn't even necessary to go to such extremes as having ten oranges on the third reel and ten plums on the second reel, as on some other Jennings machines. However, having lemons on the first reel - and on the third reel, where they are no longer used - is unfortunate, it must be admitted.

It may, however, be possible not to have any lemons, other than the one on which the 7 on the first reel is overprinted. All the melons on the second and third reels of the modified 21 Bell at the Nugget casino in Reno that Scarne wrote about were overprinted on oranges, so they could be replaced by Indian heads, and the 7 that stands on its own on the first reel could be overprinted on that reel's only lemon.

A Jennings 711 designed on that basis could have its symbols distributed this way on the reel strips:

(*) Lemons       1   0   4
(%) Cherries     2   6   0
(O) Oranges      5   3   2
(@) Plums        7   3   3
(A) Bells        1   5   8
(=) Bars         2   1   1
(4) Heads        2   2   2

this will have a higher percentage than the machine described by Scarne, since while the combinations with an overprinted melon and orange would still be present in effect, due to the payouts involving the Indian head and the orange, and the overprinting of 7s would also correspond to the overprinting of 1s on the second and third reel, there would be nothing corresponding to the overprinting of a bell and a bar on the first reel, or an overprinting of a plum and a bar on the second reel.

Rather surprisingly, it turns out that this distribution of symbols is very similar to the one I came up with above by starting with an ordinary Jennings set of reel strips.

The payouts may be calculated as follows:

Reel           Combinations  Value   Amount
 1    2    3                         Paid

 2 * 14 * 20 =          560  *   3 =  1680
 2 *  6 * 20 =          240  *   5 =  1200
 5 *  5 *  5 =          125  *  11 =  1375
 7 *  3 *  4 =           84  *  13 =  1092
 1 *  5 *  9 =           45  *  18 =   810
 2 *  1 *  1 =            2  * 100 =   200
 2 *  2 *  2 =            8  * 100 =   800
 1 *  1 *  1 =            1  * 200 =   200

giving a payout percentage of 91.96%, as opposed to 94.45% on the machine that Scarne investigated.

The Buckley Criss-Cross and Its Imitators

The first arrangement in the bottom half of the illustration is that of the Buckley Criss-Cross. Not shown in the diagram is the most important and unique feature of its payout schedule: if a bar was visible on each reel, whether on the payout line or on the line above or below, a prize of 18 coins would be paid, except of course in the case where the three bars were all on the payout line, in which case a jackpot would be awarded as usual.

The Jennings Tic-Tac-Toe came out shortly after, also incorporating this unique feature. Its payouts were slightly different; a single cherry paid 3 instead of 2, three oranges paid 11 instead of 10, but three plums paid 13 instead of 14. Its original form, as a payout option for the Jennings Standard Chief, is pictured at right. This payout style was also available with later Jennings machines, such as the Sun Chief and the Governor.

One possible arrangement of the contents of its reels are as follows:

(*) Lemons       -   -  10
(%) Cherries     1   7   -
(O) Oranges     10   1   3
(@) Plums        5   1   5
(A) Bells        1  10   1
(=) Bars         2   1   1

Making lemons fully half of the symbols on the third reel seems something of an uninspired expedient to reduce the amount paid, but the standard measure of making symbols common on one reel, and rare on another, seems to have been pushed to its limits here as well. However, in addition to allowing bars to freely replace cherries on the first two reels, this machine also offered an additional payout of eighteen coins if a bar was in any of the three visible positions for all three reels.

With its payout schedule of:

%.., =..              3
%%., %=., =%., ==.    5
OOO, OO=             11
@@@, @@=             13
AAA, AA=             18
===                 150

we can work out how much it pays:

Reel             Combinations  Value   Amount
 1    2    3                           Paid

 3 * 12 * 20 =            720  *   3 =  2160
 3 *  8 * 20 = 480 - 2 =  478  *   5 =  2390
10 *  1 *  4 =             40  *  11 =   440
 5 *  1 *  6 =             30  *  13 =   390
 1 * 10 *  2 =             20  *  18 =   360
 2 *  1 *  1 =              2  * 150 =   300

Note that while in the first instance, the number of combinations for %%., %=., =%., and ==. is calculated for all 20 symbols on the third reel, for two bars, but not the other combinations of the first two symbols, a third bar does change the winning combination to something else, so those possibilities need to be subtracted.

For the payouts given for bars in other positions, we need to add 52 times 18 coins, or 936 coins, increasing the return on 8000 coins to 6536 coins, for a profit of 18.3%.

This type of payout arrangement has an interesting history. While the Jennings Tic-Tac-Toe machine may be better known, as it was quite successful, and remained in use in casinos around the world for several years, it is not where this style of payout originated.

The Buckley Criss-Cross, pictured at right, was the first machine of this type, being advertised in late 1947.

Note that it closely resembles a Mills Silent slot machine, a line that started from the Mills War Eagle in 1931 and continued to the first post-war Mills slot machine, the Black Cherry; in particular, it strongly resembles both the Chrome Bell (Diamond Front) and the Black Cherry. This is not a coincidence: Buckley was primarily in the business of reconditioning and modifying used slot machines, and so the earliest Buckley Criss-Cross slot machines were Mills slot machines that simply had their reel strips and payout disks modified, as well as being given a new style of case. However, the popularity of the Criss-Cross was such that to meet demand for it, Buckley began manufacturing their own slot machines from scratch, and then noted that its machines were newly manufactured in its advertising.

The similar set of payouts that it offered was:

%.., =..              2
%%., %=., =%., ==.    5
OOO, OO=             10
@@@, @@=             14
AAA, AA=             18
===                 100

with three bars in other positions paying 20 coins.

The symbols on its reels were arranged as:

(*) Lemons       -   -   6
(%) Cherries     2   8   -
(O) Oranges      8   1   6
(@) Plums        8   1   5
(A) Bells        1   9   2
(=) Bars         1   1   1

which arrangement is very similar to that of the related Jennings machine.

Again, we can work out how much it pays:

Reel             Combinations  Value   Amount
 1    2    3                           Paid

 3 * 11 * 20 =            660  *   2 =  1320
 3 *  9 * 20 = 540 - 1 =  539  *   5 =  2695
 8 *  1 *  7 =             56  *  10 =   560
 8 *  1 *  6 =             48  *  14 =   672
 1 *  9 *  3 =             27  *  18 =   486
 1 *  1 *  1 =              1  * 200 =   200

and for the payouts on bars in other positions, now we need to add 26 times 20 coins, or 520 coins. That gives a payout of 5453 coins out of 8000, for a percentage of 31.8375%.

The Jennings Tic Tac Toe slot machine appears to have only come out in 1948, despite some sources ascribing a 1947 date to it. Machines of that arrangement continued to be made by them when Jennings updated the styling of its slot machines in 1949. By that time, both Pace and Watling were also offering similar payout arrangements for their machines as an option, as their advertising noted, although some advertisements by Pace, mentioning this, also recommended against it, for whatever reason. When I wrote that, I had seen advertisements that simply said a criss-cross arrangement was available, but noted "(not desirable)" afterwards; an earlier advertisement, from 1948, shown on the left, makes their reasons explicit; apparently for some reason they couldn't work out how to design reel strips that would remain profitable with such generous payouts (although they claimed to have copied those of their competitors exactly).

One advertisement for a Watling machine with this feature showed 12 extra payouts it provided: for bars in any position on the middle reels, and in the two positions other than the payline for the first and third reels. The relevant portion of the advertisement is shown to the right.

One early advertisement for the Jennings Tic-Tac-Toe, which appeared in the March 6, 1948 issue of Billboard, stated that there were twelve possibilities, rather than the full 26 which the Buckley Criss-Cross did explicitly claim to offer (for example, a full-page ad in the March 13, 1948 issue of Billboard stated "Three bars showing in any position pay 20 coins"), that paid out an additional prize of 18 coins. Looking at the design of the card at the top of the machine bearing its name, I suspect that it paid out on the same combinations as were shown in the diagram for the Watling machine.

Incidentally, note that the jackpot window in the Buckley Criss-Cross is covered, with a notice saying "Guaranteed $50.00 Jackpot" (200 coins, as it is a quarter machine). I had been looking for a more complicated answer as to why they did not have to resort to excluding some combinations in order not to pay the prize for bars in any position and a jackpot at the same time (although many older Mills machines paid the jackpot plus the old 20 coins for three bars, so this does not appear to have been mechanically impossible); but if the mechanism is not involved in paying out the jackpot at all, as it is paid manually, Gold Award style (and a completely separate mechanism is used to pay out a Gold Award token, so that mechanism could be used without interference) then there is no problem to be dealt with.

As some later Jennings Tic Tac Toe machines stated explicitly that three bars in any position paid 18 coins, I presume that they did eventually overcome the limitation of their original Tic Tac Toe mechanism.

Other Slot Machines

The second arrangement shown on the second line of the diagram is that of the Jennings Sweepstake Chief from 1950, pictured at right. This machine has the style of case introduced in 1949 for the Jennings Sun Chief. Here, the pear was added as an extra symbol, ranking below the orange. Two plums and two bells gave the usual payouts for three plums and three bells, because on this machine, three plums and three bells were alternative jackpots. As well, not shown in the diagram, three horses would pay out 500 coins if they came up when a light on the machine was lit.

How can it sustain such a generous payout schedule? Obviously, the extra pear symbol helps. A set of reel strips for the machine had the following distribution of symbols:

(%) Cherries     3   7   -
(&) Pears        6   1   1
(O) Oranges      7   8   1
(@) Plums        1   1   8
(A) Bells        1   1   8
(=) Bars         1   1   1
(^) Horses       1   1   1

However, the payout schedule wasn't actually quite as generous as the one shown in the diagram. The plums on the first two reels, and one of the plums on the third reel, were overprinted with the number 11, and, similarly, the bells on the first two reels, and one of the bells on the third reel, were overprinted with the number 7. Jackpots were not paid on any three plums or any three bells, but only on three 11s or three 7s.

Attempting to calculate its percentage:

Reel           Combinations  Value   Amount
 1    2    3                         Paid

 3 * 13 * 20 =          780  *   3 =  2340
 3 *  7 * 20 =          420  *   5 =  2100
 6 *  1 *  2 =           12  *   5 =    60
 7 *  8 *  2 =          112  *  11 =  1232
 1 *  1 * 19 =           19  *  13 =   247
 1 *  1 * 19 =           19  *  18 =   342
 1 *  1 *  1 =            1  * 100 =   100
 1 *  1 *  1 =            1  * 100 =   100
 1 *  1 *  1 =            1  * 100 =   100
 1 *  1 *  1 =            1  * 100 =   100

So even assuming a generous jackpot amount of 100 coins, it pays out 6721/8000 coins, or 84.0125%.

I didn't attempt to include the 500 coin jackpot, since I hav eno way of knowing how often the light which increased the prize for three horses would be lit.

The third arrangement shown in the bottom half of the illustration finally shows an example of payouts for a more modern type of machine, on which an award is made for three cherries in a row.

The arrangement of symbols on the reels of this slot machine, also by Jennings, is:

(%) Cherries     3   7   3
(O) Oranges      7   1  10
(@) Plums        6   1   5
(A) Bells        1  10   1
(=) Bars         3   1   1

and the payment schedule it had was the following:

%..              3
%%.              5
%%%             11
OOO, OO=        11
@@@, @@=        13
AAA, AA=        18
===            150

Once again, we can attempt to calculate this machine's percentage:

Reel           Combinations  Value   Amount
 1    2    3                         Paid

 3 * 13 * 20 =          780  *   3 =  2340
 3 *  7 * 17 =          357  *   5 =  1785
 3 *  7 *  3 =           63  *  11 =   693
 7 *  1 * 11 =           77  *  11 =   847
 6 *  1 *  6 =           36  *  13 =   468
 1 * 10 *  2 =           20  *  18 =   360
 3 *  1 *  1 =            3  * 150 =   450

for a percentage of 13.2125%.

Some machines use an upside-down horseshoe (as is favored to retain the good luck) instead of a melon, and Bally has used a target symbol on some of their slot machines as a jackpot symbol. As well, they made an early slot machine, the Bally Double Bell, where three lemons paid out 6 coins.

A slot machine does not necessarily have to have three reels.

The fourth arrangement on the second row shows the payouts for the Vendet Midget from 1931, and the later Burtmeier Pony from 1934. In order to allow only two reels to produce combinations that had a low enough probability to have payouts comparable to those of a three-reel machine, the two reels each had 38 positions on them instead of the normal 20.

The two machines were nearly identical, thus the Burtmeier Pony was likely the Vendet Midget then sold under another name. Their payout schedule, when used with fruit symbols (reel strips with images of dice in different colors were also available) was:

%%   2
OO   5
@@  10
AA  15
==  20 + Jackpot

To compensate for having two reels instead of three, it had 38 symbols on each reel instead of 20, allowing 1,444 combinations. While this was less than the 8000 combinations of a conventional three-reel machine, it was more than the 1000 combinations of a three-reel machine with 10 symbols on each reel, so it was adequate.

Interestingly enough, the two reel strips used on this machine were nearly identical; it was not necessary to compensate for having only two reels by making some symbols rare on one reel, and the others rare on the other, in order to reduce the frequency of payouts.

One set of reel strips for the machine was like this:

Lemons      8    8
Cherries   14   14
Oranges     8    8
Plums       4    4
Bells       2    2
Bars        1    1

and so it paid out like this:

14 * 14 = 196   *   2 =   392
 8 *  8 =  64   *   5 =   320
 4 *  4 =  16   *  10 =   160
 2 *  2 =   4   *  15 =    60
 1 *  1 =   1   * 120 =   120

and thus, even if the jackpot was 100 coins (20 coins were paid in addition to the jackpot), the machine would have paid 1052 coins out of 1444, for a percentage of 27.147%.

It seems bizarre that they chose to make the two reels identical, at the cost of having so many lemons on each reel. The percentage of the machine could have been exactly the same, while reducing the number of lemons, with a trivial alteration:

Lemons      1    1
Cherries   14   14
Oranges     4   16
Plums      16    1
Bells       1    4
Bars        1    1

The version with strips with pictures of dice on them was interesting. The correspondence between slot machine fruit symbols and dice was:

          Color             1  2
          -----------       -  -
Bar       Dark Purple       6  5
Bell      Brown             5  6
Plum      Blue              4  3
Orange    Green             3  4
Cherry    Mauve             2  2
Lemon     Yellow            1  1

so the winning combinations were throws of 11, 7, and 4 the hard way; the colors of the dice shown on the two reels matched, but the numbers on the dice on the two reels with the same color often differed so that the totals on the dice would be appropriate.

Speaking of Dice

A slot machine very highly prized by collectors is the Buckley Bones. This machine has two reels, but instead of the reels having paper strips on which symbols are printed, the reels have compartments, each containing two dice. So as the reels spin to present a compartment with a pair of dice within it, the machine presents an illusion of throwing dice.

The Buckley Bones and the Bally Reliance, except for their branding and styling, are essentially identical. Both first became available in 1936. The Bally Reliance is pictured at right.

The operation of these machines follows the course of the game of craps.

After inserting a coin, pulling the lever spins the first reel only. If a compartment shows with 2, 3, or 12, that is the end of the player's turn, and the player loses.

If a compartment shows with 7 or 11, the player wins 2 coins.

Otherwise, without inserting another coin, the player may pull the lever again, this time spinning the second reel. If a 7 turns up, the player loses. If the number shown on the first reel is matched, the player wins; but instead of winning even money, as in craps, the payoff varies depending on wht the pont is.

If the player's point is 6 or 8, winning pays even money. If the point is 4, 5, 9, or 10, however, a winning player is paid 8 for 1 (or 7 to 1).

Also, if one won by an initial 7 or 11 four times in a row, one won a Gold Award, presumably a jackpot-sized prize.

Each reel on the Reliance or the Bones had 13 compartments.

The contents of the left reel were:

 7  1
11  1

 2  1
 3  -
12  1

 4  1
 5  1
 6  2
 8  3
 9  1
10  1

So initially the player had a 2/13 chance of losing and a 2/13 chance of winning. The chance of a Gold Award was 16/28561.

The contents of the right reel were:

 7  7

 4  1
 5  1
 6  1
 8  1
 9  1
10  1

If a point other than the player's point comes up, the player can pull the lever again without paying another coin for another spin of the second reel.

Thus we can calculate the machine's payout percentage: out of 104 coins played, 13 times 8, the player will recieve:

2/13 * 2 coins                 32 (out of 104)

4/13 * 1/8 * 8 coins           32

2/13 * 1/8 * 2 coins            4

16/28561 * 100 coins            4.826
                               72.826 out of 104

for a payout percentage of 70.025%, as the Gold Award was worth 100 coins.

The Mills Novelty Company made a machine of their own which competed with the Reliance and the Bones which was even more elaborate, the Mills Dice. It may even have predated the Reliance and the Bones, for that matter, as it seems to have been advertised a month earlier than they were.

Before either the Mills Dice and the Reliance and the Bones, Western Equipment and Supply had a very successful machine, with an automatic payout, that also was designed to give the appearance of dice being thrown, the Mysterious Eye. It was simpler than even the Reliance, one spin simply led to two pairs of dice becoming immediately visible - and the winning combinations were not closely patterned after those of the game of craps either. It was popular at the time, setting sales records in 1935, so it is would not have been surprising that it inspired attempts to improve on it.

However, I learned that Mills had applied for a patent on the Mills Dice machine fairly early in its development process, so it was actually that patent that created a buzz in the amusement machine industry and led to the development of both the Reliance and the Mysterious Eye.

The Mysterious Eye is described in U. S. Patent 2,135,182. It turns out that it managed to provide automatic payout while actually throwing dice. This was managed by using dice that were not perfectly cubical, but which instead had at least one pair of faces slightly closer together than the others. Thus, after the dice were thrown, the baseplate moved up to hold the dice against the glass faceplate, and partial information about how the dice were facing could be obtained from where the base plate moved.

Assuming the alteration in the shape of the dice did not materially affect how they rolled:

The machine awarded "10 points" for any combination of 1 and 6 in the first two compartments, which would have a 1/9 chance of occurring.

But it would award "30 points" instead if 1 or 6 also came up in the third compartment.

If 1, 3, 4, or 6 came up in the fourth compartment when either of the previous winning combinations was present, "30 points" would also be awarded.

So on the first three dice, 1 and 6 were the faces that were closer together, while on the fourth die, 2 and 5 were closer together.

The chance of only winning 10 points would be 1/9 times (2/3 times 1/3), or 2/81, and the chance of winning 30 points would be 7/81, if I understand the payout schedule correctly.

A 10 point win returned two nickels, so assuming a 30 point win returns six nickels, and one pays a nickel to play, the machine would return 46 nickels for every 81 nickels played, about 56.79%, for a house advantage of 43.21%.

On the Mills Dice, pictured at right, there were three openings through which a coin could be inserted for three possible bets.

One could bet on 11, for a payout of 16 coins if one won.

One could bet on the field, for a payout of 2 coins if either 2, 5, 9, 10, 11, or 12 came up; originally, I thought it was if 2, 3, 5, 8, 10, 11, or 12 came up, as that was what was shown on an early advertisement for the machine, but I have since learned the correct set of numbers for the field, first from a forum post, and then I checked a video of an actual machine to confirm this. The larger set of winning numbers shown in the advertisement would have had an exact 50% chance of coming up, which would mean that bet would make no profit.

One could play a pass/come bet, which followed the normal rules of the game of Craps; 7 or 11 won, 2, 3, or 12 lost, on the first throw, and subsequently one won with one's point, or lost with a 7.

The machine presented a more elaborate illusion of dice being thrown than the Reliance and Bones.

Two dice were ejected, from a ring of thirty dice, into a round area. The area was divided into two chambers, one for each die. In preparation for a subsequent throw, each die would be returned from its chamber into the part of the ring from which it came. The dice had no opportunity to tumble; the numbers on the die were determined by the part of the ring of dice from which they came, not by tumbling when thrown.

All throws of the dice, the first one and subsequent ones, came from the same thirty possibilities, thus making this machine different from the Reliance and Bones, which had one reel for the come-out roll and one reel for subsequent rolls.

The thirty possible rolls of the dice were distributed as follows:

 7   5
11   1

 2   2
 3   4
12   2

 4   2
 5   2
 6   4
 8   4
 9   2
10   2

A throw of 7 comes up 5 times out of 30, one-sixth of the time, exactly as with real dice. Other possibilities, however, have changed.

11 comes up one time out of 30, but only pays 16 coins, so that bet only returns 53.33% of what the player wagers.

The winning Field numbers are 2, 5, 9, 10, 11, or 12, and those six numbers cover 11 possibilities, returning 22 coins out of 30, or 73.333...%. Had the numbers I had seen in the early advertisement been the winners instead, 2, 3, 5, 8, 10, 11, or 12, they would cover 15 out of 30 possibilities, making that bet fair.

The bet that is presumably the most popular would be the one involving most of the machine's complexity, the one that simulates the game of craps.

The odds break down like this:

6/30 chance of winning on the first roll (7,11)                756 coins
8/30 chance of losing on the first roll (2, 3, 12)
8/30 chance of having a 2/7 chance to win (4, 5, 9, 10)        288 coins
8/30 chance of having a 4/9 chance to win (6, 8)               448 coins

so out of 63 * 30 coins played, that is, 1,890 coins, the player would recieve 1.492 coins in return, for a payout percentage of 78.94%, and thus that bet yielded a house advantage of 21.06%.

A Historical Note on the Bally Reliance

When Bally introuduced the Reliance at the 1936 Coin Machine Show, this introduction was accompanied by a show, "Dance of the Dice", presented by The Two Zephyrs, a comedy and dance act.

The Two Zephyrs performed with Duke Ellington over a period of several years, so they were apparently an act of some significance.

One of the Two Zephyrs, who I only know of as "Big Zeph", passed away in 1940, and Melvin Edward White, known as "Slappy" White, replaced him. The remaining partner, "Little Zeph", was Clarence Schelle.

Melvin White later left the Two Zephyrs; another comedy act, Canfield and Lewis, consisting of James "Spizzy" Canfield and Willie Lewis broke up, and White joined with Lewis to form Lewis and White. Then, in 1947, he moved on again, to team up with Redd Foxx.

Of course, we remember Redd Foxx from the television show "Sanford and Son", on which Melvin Lewis made a guest appearance; before, he had a... salty... comedy routine, and some phonograph records of it exist as well.

Meanwhile, the Two Zephyrs continued to exist, performing with Duke Ellington at least as late as 1950 (and as early as 1937, as well, from accounts I have seen). So the act did continue after Melvin Lewis, replacing him.

However, while Melvin Lewis is remembered due to other successes in his career, so far I have not been able to find out who he replaced, or who he was replaced by, in the few accounts of the group I have been able to find online.

And Now for Four Reels

The fifth arrangement is that of another machine made by Jennings. The payout for four plums is depicted as 26; on the original Jennings Buckaroo, it was 22 in order to simplify the payout mechanism slightly, but on the Jennings Golden Nugget, it was made 26 to be double that for three plums. The Jennings Buckaroo was the first four-reel machine introduced after World War II. The most common date I have seen given for it is 1955, but some sources say it came out in 1958, and one source says that Jennings at least began producing them in 1954. This is an important date because in advertising for Jennings slot machines up to 1950, when slot machines were still being widely sold, no evidence of cherries on the third reel has turned up; since the payout structure of the four-reel Jennings Buckaroo includes cherries on the third reel, if not the fourth, it seems likely that their three-reel machines with cherries on the third reel either became available at the same time, or preceded it.

One arrangement of symbols on the reels for the Jennings Buckaroo (that of the V12-117 reel strips) is the following:

Cherry         2  6  3  -
Orange         6  2  4 10
Plum           4  1 10  6
Bell           3  7  1  3
Bar            4  3  1  -
Buckaroo       1  1  1  1

which leads to payouts (from 160,000 coins instead of 8,000) as follows:

             Reel              Combinations  Value        Amount
              1    2    3    4                              Paid

%...     3    2 * 14 * 20 * 20 =     11,200  *     3   =  33,600
%%..     5    2 *  6 * 17 * 20 =      4,080  *     5   =  20,400
%%%.    11    2 *  6 *  3 * 20 =        720  *    11   =   7,920
OOO.    11    6 *  2 *  4 * 10 =        480  *    11   =   5,280
OOOO    22    6 *  2 *  4 * 10 =        480  *    22   =  10,560
@@@.    13    4 *  1 * 10 * 14 =        560  *    13   =   7,280
@@@@    26    4 *  1 * 10 *  6 =        240  *    26   =   6,240
AAA.    18    3 *  7 *  1 * 17 =        357  *    18   =   6,426
AAAA   100    3 *  7 *  1 *  3 =         63  *   100   =   6,300
===.   100    4 *  3 *  1 * 20 =        240  *   100   =  24,000
BBB.   100    1 *  1 *  1 * 19 =         19  *   100   =   1,900
BBBB  2000    1 *  1 *  1 *  1 =          1  * 2,000   =   2,000

which gives a payout percentage of just over 82.44%.

The V12-112 reel strips give another arrangement of symbols on the reels for the Jennings Buckaroo, with slight modifications to the first two reels, and a major transformation of the fourth reel. However, there are no oranges on the fourth reel in that set as pictured in my source; if I change one of the plums on that reel to an orange, I would get:

Cherry         2  7  3  -
Orange         7  2  4  1
Plum           5  1 10  6
Bell           3  8  1  3
Bar            2  1  1  -
Buckaroo       1  1  1 10

and that would work out to the following:

             Reel              Combinations  Value        Amount
              1    2    3    4                              Paid

%...     3    2 * 13 * 20 * 20 =     10,400  *     3   =  31,200
%%..     5    2 *  7 * 17 * 20 =      4,760  *     5   =  23,800
%%%.    11    2 *  7 *  3 * 20 =        840  *    11   =   9,240
OOO.    11    7 *  2 *  4 * 19 =      1,064  *    11   =  11,704
OOOO    22    7 *  2 *  4 *  1 =         56  *    22   =  10,560
@@@.    13    5 *  1 * 10 * 14 =        700  *    13   =   9,100
@@@@    26    5 *  1 * 10 *  6 =        300  *    26   =   7,800
AAA.    18    3 *  8 *  1 * 17 =        408  *    18   =   7,344
AAAA   100    3 *  8 *  1 *  3 =         72  *   100   =   7,200
===.   100    2 *  1 *  1 * 20 =         40  *   100   =   4,000
BBB.   100    1 *  1 *  1 * 10 =         10  *   100   =   1,000
BBBB  2000    1 *  1 *  1 * 10 =         10  * 2,000   =  20,000

which gives a payout percentage of just over 89.34%, so changing the plum to an orange did not create an unprofitable, and hence impossible, set of reel strips.

The payouts for the Jennings Buckaroo are also reproduced again in the second column of the image below:

The first four-reel conventional slot machine (five-reel machines with playing card symbols, and without automatic payout, predated the conventional slot machine) was one by Caille from 1930, well before the war, illustrated at right. Instead of extending the machine by adding another reel on the right, the Caille Grand Prize, a four-reel model of the Caille Superior, had one extra reel on the left; the only symbol on it was a lemon, the other positions being blank; thus, that lemon was required for a special larger jackpot, while the prizes for the combinations on the other three reels remained unaffected.

That resulted in a payout schedule like that illustrated as the first arrangement in the image above however, one could only win the grand prize if the symbols lined up also had the digits 1, 2, 3, and 4 overprinted on them, and this was not true of all the lemons on the second reel, all the bells on the third reel, or all the lemons on the fourth reel, so winning the prize was more unlikely than just having all the appropriate symbols lined up would be.

One set of reel strips for this machine had the following distribution of symbols:

Blank          16  -  -  -
Lemon (U)       -  1  -  2
Lemon (N)       4  2  -  2
Cherry          -  5  5  -
Orange          -  3  5  5
Plum            -  3  3  5
Bell (U)        -  4  3  3
Bell (N)        -  -  2  -
Bar             -  2  2  3

Thus, out of 160,000 pulls, the four-reel jackpot would come up 32 times, or once out of 5,000 pulls. This prize was typically equivalent to 100 coins, so it was neither particularly larger nor particularly more unlikely, than a jackpot could be from a conventional three-reel machine.

Although the machine had automatic payout for the other prizes, the four-reel jackpot was awarded by the operator.

The five-reel machines with playing cards that preceded Charles Fey's original Liberty Bell slot machine were classed as trade stimulators in the terminology used by antique collectors, as one had to put a coin in to pull the lever and spin the wheels, but then the merchant would have to look at the combination displayed to pay out any prize manually. The "New Deal" trade stimulator by Pierce had five reels, but one could change the front of the machine to display only three or four of them. So it could have five reels with playing cards, with prizes awarded based on poker hands, it could have three reels with the Mills fruit symbols, and payouts like those of a slot machine, or it could have four reels, where the three reels in the middle had Mills fruit symbols, but the fourth reel, either on the left or the right, showed a payout amount, which would be awarded if any three matching fruit symbols came up.

Another more conventional four-reel slot machine, in addition to the one made by Jennings, was made by Pace, as a variation of the Pace Comet Deluxe, and its payouts are illustrated in the third column of the illustration above.

The fourth column shows a set of payouts used by Mills, when, somewhat late in the era of mechanical slot machines, it presented its own take on the four-reel slot machine in a Hi-Top body.

I've also seen images of four-reel slot machines apparently by Mills with other payout schedules: one with a basically identical payout schedule, except that four oranges only pay 14 and four plums only pay 18, and one where in addition to substituting for oranges, plums, and bells on the third reel, the bar can substitute for oranges, plums, and bells on the fourth reel as well.

Instead of the melon, shown here as an additional jackpot symbol, the machine as manufactured used an owl, the Mills trademark, in a circle, but that symbol was often replaced by a symbol representing the casino using the machine.

Like the Pace four-reel machine, four oranges and four plums give the same prize, and like both the Jennings Buckaroo and the Pace four-reel machine, four bells are a jackpot.

But in some ways, this machine's payout schedule illustrates that Mills remained steeped in classic slot machine tradition. Cherries are still found only on the first two reels. The bar substitutes for an orange, plum, or bell on the third reel, where it had always done so on three-reel machines.

More recently, slot machines had been designed to have 22, 23, or 25 positions per reel, and currently electronic machines are used in casinos that simulate a slot machine with perhaps 128 positions on each reel, and so it's hardly necessary to add a fourth reel if one wishes to have a large jackpot that would have to also be very infrequent.

Alternate Fruit Symbols

While the fruit symbols introduced on the Mills Operator's Bell quickly became the standard, slot machines were made which instead had patriotic symbols on the reels, or symbols associated with baseball or other sports.

One company, Caille, initially attempted to use fruit symbols, as made popular by the Mills machines, without copying their set of fruit symbols exactly.

The first set of payouts in this illustration show the appearance of the symbols in the Caille Operator's Bell. In the Caille Liberty Bell, which used the same fruit symbols, the letters G, U, and M overlaid the fruit symbols as in the first Mills machines with fruit symbols, where the spearmint leaf was present instead of the cherry. However, while the reel strips of those Mills machines represented the letters G, U, and M in a variety of ways, for example drawing square letters on the Bell, building letters out of many small Plums, and intertwining Lemons on a branch with the letters, Caille simply overprinted the letters in the same style on all the symbols.

Here, the bar was a stick of Liberty Fruit Chewing Gum, at a 45 degree angle, with a patriotic shield symbol.

The apple, with a stem above it, looked like it might have actually been a single cherry, but the same drawing, with "Apple" on a green bar across it, appeared in the later Caille Victory Bell, so I can only accept that it was intended to be an apple.

The second set of payout symbols was used with versions of the Operator's Bell that replaced the bell symbol with another fruit, the pineapple, to further emphasize the machine's role as a gum vendor, as opposed to a gambling device.

The third set of payouts in this illustration are those of the Caille Victory Bell, which is of historical interest as the apple, strawberry and pineapple made their appearance again as three of the additional fruit symbols used when Adolph Caille founded A. C. Novelty and brought out the innovative and unusual Multi-Bell Seven-Way, to be discussed below.

Note that while the Liberty Bell remained a symbol on this machine, it took over the role of the bar, while the pineapple took over that of the bell. On the other hand, plums (as well as the lemon) kept their original role, but were drawn differently, the symbol depicting a pair of plums rather than a single plum.

The Victory Bell was made during the Second World War; or, rather, since metals were critical war materials, older machines were converted to this type with new reel strips.

Other early Caille machines used other fruit symbols in a different way:

The fourth set of payouts are those of the Caille Little Merchant, as well as the nearly-identical Watling O.K. Deluxe. Note that here, bananas replace the cherries, pineapples replace oranges (instead of bells), pears replace the plums, grapes replace the bells, and strawberries replace bars (instead of oranges). The lemon retains its identity, but is drawn differently, at an angle and on a branch.

And on one occasion, Caille took the standard Mills fruit symbols, and had a bit of fun with them, surrounding each symbol with a dynamic-looking frame. This is illustrated in the fifth set of payouts, belonging to a machine from 1934 which was no doubt terribly embarassing in hindsight, the Caille Dictator.

The sixth set of payouts are those of a smaller machine with a wooden case, the Caille Merchant. In addition to avoiding confusion between the Caille Little Merchant and the Caille Merchant, there was also a much later small machine by Jennings called the Little Merchant. Note how the bar is now drawn as a stick of gum at an angle, depicted as a three-dimensional object.

A number of books about slot machines, or about gambling in general, refer to a persistent myth that it is possible for operators to just turn a screw somewhere in a slot machine to adjust its payout percentage. Instead, they note, it would be necessary on a mechanical slot machine to change the symbols on the reel strips, and also change the disks with holes in them which indicate to the mechanism which symbol is on each reel in a given position in a corresponding manner.

One might be able to change a machine's percentage slightly with a screwdriver by changing how large the maximum size of the jackpot is allowed to be.

That sounds very sensible to me, particularly as those accounts did include the qualification that something like that might be possible if the machine was rigged.

However, I have encountered what may be the source of this rumor.

An early Caille catalog, in its entry for the Caille Liberty Bell, contains the following statement:

"Percentage regulator changes profits from 25% to 50% without change of reward card."

While the machine could have come with an alternate set of reel strips, this certainly does sound like the kind of adjustment that was rumored.

However, I have found more information in a book about the 1911 session of the California legislature. At that time, both Mills and Caille upright machines had in them a two-position lever which, when set in the higher percentage position, prevented the rarer colors with higher payouts from coming up when they were played, but they could still come up if the colors at lower odds were palyed instead. So appareently the "percentage regulator" on a slot machine would make it behave as if it were "bugged".

A Different Machine

Before turning to the chief topic of this section, to introduce the ideas it uses, I would like to examine another unusual slot machine that is less well known, the Alwin by Buckley, also sold as the Deluxe Cent-a-Pack. (There was also a plain Cent-a-Pack which was a normal three-reeler.)

This machine had four reels.

The first reel had a number indicating a prize amount, such as 4, 5, 10, 15, 20 or 30.

The remaining three reels had the symbols of a conventional slot machine, such as cherries, oranges, lemons and bells.

If three of any symbol showed on the pay line, one received the payout amount shown on the first reel, except that three bars still paid a jackpot of 50.

There was also a four-reel machine called "The New Deal" by Pace which may have been similar; some of those machines had the payout amount on the first reel like the Alwin, and others had it on the fourth reel. (There was also a three-reel conventional slot machine version, and a five-reel one with playing card images on the reels, which paid off for poker hands, so they used this name for machines of different kinds.)

An unusual slot machine, called the Multi-Bell Seven Way was devised by Adolph Caille after he had sold the Caille Brothers slot machine business, and so he started a new company, A. C. Novelty, to produce and sell it. (Eventually, the company was renamed to A. C. Manufacturing during the 1950s.) The machine is pictured on the left.

It had twelve different kinds of symbols; the seven which could form a jackpot combination, and on which the player could wager: Strawberries, Bells, Lemons, Bars, Plums, Apples, and Oranges, and five others: Cherries, Pears, Apricots, Grapes, and Pineapples, which only appeared on the second and third reels.

The player could choose which one of the first seven symbols to wager on, and would win a prize if that symbol came up on the first reel. The amount of the prize was indicated on the last reel; it might be anything from twice to 20 times the amount wagered.

If three of the symbol chosen by the player came up, then a jackpot would be paid out.

The payout odds were given on the third reel along with the third symbol; those odds were not the same for every occurrence of the same symbol.

The diagram to the right shows the arrangement of the symbols on its reel strips as I had reconstructed it from images of the machine, as well as from a couple of videos of the machine I had encountered.

The videos were of the version with larger numbers on the third reel; later, I saw an image of the actual reel strips for the version with smaller numbers, which confirmed my reconstruction of the first two reels, which apparently are the same in both versions, and so now the image at right shows both versions of the strip for the third reel.

The second reel had two copies of each of those seven symbols printed on their reel strips, as well as seven other symbols, and the third reel only one of each of those symbols: even so, in order for jackpots to be rare, it was still necessary that only three of the chosen symbol could pay a jackpot. On the second reel, the extra symbols were one bunch or grapes, one pear, two pineapples, and three apricots, according to one source I found; but this source also claims, incorrectly, that cherries do not appear on the third reel.

While three of any of the seven fruit symbols on which the player could wager would constitute one of the seven jackpot combinations offered by the machine, players were only eligible for the jackpots corresponding to the symbol, or symbols, on which they wagered. It was still necessary to reduce the frequency of jackpot combinations, and so the middle reel operated as a 14-stop reel; that is, one out of every three positions printed on the reel strip could not actually appear on the pay line.

It is not surprising, unfortunately, that this machine was not a great success at the time. It was ungainly and unattractive in appearance, while conventional slot machines were competing to attract the eye. More important, its play was both unfamiliar and complicated. Due to its uniqueness and rarity, however, it is highly sought after by collectors today.

This machine hearkened back to the early days of coin-operated gambling devices before Charles Fey invented the slot machine as we know it today. Basically, a typical upright was a coin-operated version of a Wheel of Fortune. The player can choose to wager on a color that only appears a few times on the wheel, to win a large prize if it comes up, or on colors that appear more often for lower prizes.

One of the earlier such machines was the Mills Owl, pictured at right, with the picture retouched by hand to make the colors of the segments of the wheel visible. It was a successor to their earlier Klondyke from 1897.

The Mills Owl had a wheel with 50 colored segments, in five different colors.

The breakdown of colors and their payouts were:

Red       18    2
Black     18    2
Yellow     8    5
Gray       4   10
Green      2   20

The first column shows the color, the second the number of spaces on the wheel that are that color, and the third the payout in coins if that color came up and was the color on which the player bet by inserting a coin in the associated slot.

An even-money bet on red or black would return 36 coins out of every 50, for a house advantage of 28%.

The other bets return 40 coins out of every 50, for (surprisingly) a lower house advantage of 20%.

The green spaces on the wheel, a dark green in color, also had an owl symbol on them, which would later become the Mills trademark.

Historical Perspective

One thing I haven't done much so far is to provide an overview of the history of slot machines as a whole as an enterprise; this page started with a focus on how the payout schedules of slot machines evolved, and not on other aspects of the story.

After Charles Fey invented the first three-reel automatic payout slot machine in San Francisco, the 1906 San Francisco earthquake prevented him from gradually, with his new invention, acquiring the ability to establish a major slot machine manufacturer.

Mills, Caille, and Watling were, at the time, all established manufacturers of upright coin-operated gambling machines like the Mills Owl shown above. Mills was the first company to begin making the new slot machines that Charles Fey invented; apparently, this was with Charles Fey's direct assistance in return for a supply of the initial production to sell and re-establish himself in business.

Caille and Watling later also entered the field, and despite competition between those companies, apparently the technology was shared on amicable terms. Also, for a time, Charles Fey was employed by the Watling firm.

O. D. Jennings was a Mills salesman, and Ed Pace had a business selling used slot machines, before they entered into the manufacture of slot machines, founding two more of the early firms in the business. Other firms, like Buckley, had similar origins much later on.

As may be apparent from what is present elsewhere on these pages, Mills was the largest company in the slot machine business, and was responsible for numerous innovations.

Caille, Jennings, and Pace came out with a few innovations of their own, and generally tried to keep up with Mills in features and styling.

After making what was perhaps the most beautiful slot machine ever made, the Rol-a-Top, Watling, on the other hand, while it did keep up with changes to payout schedules and other features, like "future pay", decided it couldn't really improve on perfection. And so, while there were the Diamond Front and Castle Front versions of the Rol-a-Top, it largely didn't change over the years from 1935 to 1950, and Watling was significantly behind the other major slot machine companies in sales.

However beautiful a slot machine may look, the passing customer will be more confident that a modern-looking one will actually work when he puts his coin in the slot and pulls the lever. And so shiny chrome machines from Pace and Jennings that were plain in comparison pulled ahead.

It has been noted above that the Caille brothers sold out to retire, and then returned to the field with a very innovative machine that was not sufficiently successful to permit the founding of a major new slot machine producer, although it seems to have enjoyed at least a modest degree of success, sufficient to permit a few variations on the design to be produced.

Mills and Jennings attempted to stay in the field when casinos switched to using electromechanical slot machines, but they were not able to do so, leaving Bally unchallenged. But then Bally in its turn succumbed to other companies achieving dominance when the switch was made to fully electronic slots, either using reels as indicators only, or with video displays.

A Stateful Machine

The Mills Futurity slot machine, pictured at right, is popular with slot machine collectors. It had the feature that if one went ten consecutive plays without a win, one would get one's ten coins back. Since some wins paid less than ten coins, it was still possible for the machine to make a profit.

One way to help that along would simply be to make the lower payments particularly common. However, the Mills Futurity went further than that; it was a "ten-stop" machine of a very special kind.

That is, although there were twenty symbols printed on each wheel, the wheels could actually only stop on ten of them.

Ten-stop slot machines were a common form of crooked slot machine in the old days. However, the Mills Futurity had a difference: depending on where the indicator of consecutive non-winning plays pointed, a different set of ten symbols was available.

The Pace Kitty, a machine inspired by the Mills Futurity, and intended to compete with it, did not resort to any such technique; however, it funded the jackpots which it paid out after ten losing plays by putting "kitty" symbols on the first reel which could not be part of any winning combination.

The Mills Futurity worked like this:

                Usual set of                     Alternate set
                symbols                          of symbols (4 or 10 losing plays)

(*) Lemons       -   -   3                       -   -   -
(%) Cherries     1   1   -                       8   8   -
(O) Oranges      3   3   1                       -   -   5
(@) Plums        2   1   1                       -   -   3
(A) Bells        0   5   4                       1   1   1
(=) Bars         4   0   1                       1   1   1

Another variation of the Mills Futurity existed with the Gold Award symbol as well: and one version of the reel strips for that machine are like this:

                Usual set of                     Alternate set
                symbols                          of symbols (5 or 10 losing plays)

(*) Lemons       1   -   3                       -   -   1
(%) Cherries     1   1   -                       8   8   -
(O) Oranges      1   3   1                       -   -   4
(@) Plums        2   1   1                       -   -   3
(A) Bells        1   1   1                       -   -   -
(=) Bars         4   0   1                       1   1   1
(o) Award        -   4   3                       1   1   1

On pulls 4 and 10, the winning combinations of three bells and the jackpot, three bars, were at least possible, so one could consider the machine to not actually be rigged in an outright fashion. Incidentally, the cheating mechanism could be removed from the machine, and so it may be useful to examine the result of doing so.

The payout odds for this machine, for the different kinds of pull, are shown below:

               Usual                       Pulls 4 and 10              Combined

%%.        3   1 * 1 * 3  *   3 =    9     8 * 8 * 9  *   3 = 1728      9 *  9 * 12  *   3 = 2916
%%*, %%A   5   1 * 1 * 7  *   5 =   35     8 * 8 * 1  *   5 =  320      9 *  9 *  8  *   5 = 3240 
OOO, OO=  10   3 * 3 * 2  *  10 =  180                                  3 *  3 *  8  *  10 =  720
@@@, @@=  14   2 * 1 * 2  *  14 =   56                                  2 *  1 *  5  *  14 =  140
AAA, AA=  18                               1 * 1 * 2  *  18 =   36      1 *  6 *  6  *  18 =  648 
===       20                               1 * 1 * 1  *  20 =   20      5 *  1 *  2  *  20 =  200
                                   ---                        ----                           ----
                                   280                        2104                           7854

with three bars also paying off the jackpot.

So normally it only pays 28% of what it takes in, but on the fourth and tenth spins, it pays out more than twice what it takes in.

In the case of the Gold Award machine, the payouts are:

               Usual                       Pulls 5 and 10              Combined

%%.        3   1 * 1 * 6  *   3 =   18     8 * 8 * 9  *   3 = 1728      9 *  9 * 15  *   3 = 3645
%%*, %%A   5   1 * 1 * 4  *   5 =   20     8 * 8 * 1  *   5 =  320      9 *  9 *  5  *   5 = 2025 
OOO, OO=  10   1 * 3 * 2  *  10 =   60                                  1 *  3 *  7  *  10 =  210
@@@, @@=  14   2 * 1 * 2  *  14 =   56                                  2 *  1 *  6  *  14 =  168
AAA, AA=  18   1 * 1 * 2  *  18 =   36                                  1 *  1 *  3  *  18 =   54 
===       20                               1 * 1 * 1  *  20 =   20      5 *  1 *  2  *  20 =  200
ooo       50                               1 * 1 * 1  *  50 =   50      1 *  1 *  1  *  50 =   50
                                   ---                        ----                           ----
                                   190                        2118                           6352

Three bars also pays the jackpot. The Gold Award does not pay a jackpot, but is instead a fixed prize that is paid in the form of a "Gold Award" token to be redeemed with the merchant. That prize might be 20, 50, 75, or 100 coins, depending on the machine. The calculation above was made assuming the Gold Award prize to be 50 coins; however, I have found that reproduction award cards stating the Gold Award to be 75 coins appear to be the ones currently most offered for sale.

If the cheating mechanism is taken out, then, in the absence of the Futurity feature, and not considering jackpots, the machine would be profitable, but it would have a low percentage in the case of the basic machine; in the case of the Gold Award machine, however, the percentage would be a more usual one of 20.6% in the absence of jackpots.

To figure out the actual percentage of the machine, one has to follow the sequence of events from the first pull. To keep the numbers reasonable, I will scale things to 1000 plays at the beginning, allowing fractional values for the additional spins, rather than either using more decimals and stating things in terms of a single play, or ensuring all numbers are integers by counting over 10 to the 30th power sequences.

Usual:        1000 combinations,  32 winning, 968 losing
Pulls 4 & 10: 1000 combinations, 643 winning, 357 losing

Thus, one has:


  1)                             32 chances of an average win of  280/32,  on an expenditure of  1 coin
  2)              968/1000    *  32 chances of an average win of  280/32,  on an expenditure of  2 coins
  3)             (968/1000)^2 *  32 chances of an average win of  280/32,  on an expenditure of  3 coins
  4)             (968/1000)^3 * 643 chances of an average win of 2104/643, on an expenditure of  4 coins
  5)  357/1000 * (968/1000)^3 *  32 chances of an average win of  280/32,  on an expenditure of  5 coins
  6)  357/1000 * (968/1000)^4 *  32 chances of an average win of  280/32,  on an expenditure of  6 coins
  7)  357/1000 * (968/1000)^5 *  32 chances of an average win of  280/32,  on an expenditure of  7 coins
  8)  357/1000 * (968/1000)^6 *  32 chances of an average win of  280/32,  on an expenditure of  8 coins
  9)  357/1000 * (968/1000)^7 *  32 chances of an average win of  280/32,  on an expenditure of  9 coins
 10)  357/1000 * (968/1000)^8 * 643 chances of an average win of 2104/643,  on an expenditure of 10 coins
      357/1000 * (968/1000)^8 * 357 chances of the return of the 10 coins played

Thus, the percentage for one complete series of pulls, long or short, until a prize is returned can be worked out:


         Expenditure          Return
 1)      1000                    280
 2)       968                    271.04
 3)       937.024                262.36672
 4)       907.039232            1908.410544
 5)       323.813006              90.667642
 6)       313.450990              87.766277
 7)       303.420558              84.957756
 8)       293.711100              82.239108
 9)       284.312345              79.607457
10)       275.214350             579.050992
         -----------            -----------
         5605.985581            4708.621725

Since by that time, one has spent almost 4,000 coins, quitting after the fourth pull, which returns half of them, is not profitable, although it's less bad than quitting after any of the first three pulls. Similarly, if one goes past the fourth pull, it pays to continue until the tenth pull.

Thus, exclusive of jackpots, it pays out 83.99% of what it takes in, for a percentage of 16.01%, which is not bad for a slot machine that is "rigged". And this does not even count any jackpot paid in addition to the 18 coins for three bars.

Of course, the machine does pay out jackpots. If we assume that the jackpot consists of 100 coins, the number of coins paid for jackpots - which can only happen on the fourth and tenth plays - can be calculated in the same fashion:


  4)             (968/1000)^3 *  1 chance of a win of 100
 10)  357/1000 * (968/1000)^8 *  1 chance of a win of 100

giving the following increment on the return:


 4)                               90.703923
10)                               27.521435

This amounts to 2.11% of what was spent, so the payout is increased to 86.1%, reducing the machine's percentage to 13.9%. For other values of the jackpot, the change to the percentage scales linearly, as the jackpot amount does not affect which spins win.

From one source, it appears that the Gold Award version of this machine switched to the alternate symbols on the fifth and tenth spins instead of the fourth and tenth spins. This could have the effect of increasing the percentage of the machine, at least if otherwise it would be worthwhile to quit after the fourth spin: but since it has a higher percentage for the house on the normal spins, it is unclear why this change would be needed.

As I'm not aware that anyone else has attempted to calculate the percentage of a Mills Futurity slot machine, I've decided that it's probably worthwhile to attempt to calculate the percentage for the Gold Award version of it as well.

So, repeating the calculation:

Usual:        1000 combinations,  22 winning, 978 losing
Pulls 5 & 10: 1000 combinations, 642 winning, 358 losing

Thus, one has:


  1)                             22 chances of an average win of  190/22,  on an expenditure of  1 coin
  2)              978/1000    *  22 chances of an average win of  190/22,  on an expenditure of  2 coins
  3)             (978/1000)^2 *  22 chances of an average win of  190/22,  on an expenditure of  3 coins
  4)             (978/1000)^3 *  22 chances of an average win of  190/22,  on an expenditure of  4 coins
  5)             (978/1000)^4 * 642 chances of an average win of 2118/642, on an expenditure of  5 coins
  6)  358/1000 * (978/1000)^4 *  22 chances of an average win of  190/22,  on an expenditure of  6 coins
  7)  358/1000 * (978/1000)^5 *  22 chances of an average win of  190/22,  on an expenditure of  7 coins
  8)  358/1000 * (978/1000)^6 *  22 chances of an average win of  190/22,  on an expenditure of  8 coins
  9)  358/1000 * (978/1000)^7 *  22 chances of an average win of  190/22,  on an expenditure of  9 coins
 10)  358/1000 * (978/1000)^8 * 642 chances of an average win of 2118/642, on an expenditure of 10 coins
      358/1000 * (978/1000)^8 * 358 chances of the return of the 10 coins played

Thus, the percentage for one complete series of pulls, long or short, until a prize is returned can be worked out:


         Expenditure          Return
 1)      1000                    190
 2)       978                    185.82
 3)       956.484                181.73196
 4)       935.441352             177.733857
 5)       914.861642            1937.676958
 6)       327.520468              62.228889
 7)       320.315018              60.859853
 8)       313.268087              59.520937
 9)       306.376189              58.211476
10)       299.635913             634.628864
         -----------            -----------
         6351.902670            4621.109363

So in this case, the machine returns 72.75% of what is put in, for a profit of 27.25%, which, again, is highly reasonable for a slot machine. This is based on the assumption that the Gold Award payout is 50 coins, but excludes any jackpot paid on the three bars. It is, however, a higher percentage, as anticipated, than that of the plain version.

As both three Gold Award symbols and three bars occur on only one combination, and only for the fifth and tenth spins, the same calculation will let us determine how to adjust the percentage for the amount of the jackpot and for a change in the size of the Gold Award from the value of 50 that was assumed.

Again, for a jackpot of 100:


  5)             (978/1000)^4 *  1 chance of a win of 100
 10)  358/1000 * (978/1000)^8 *  1 chance of a win of 100

giving the following increment on the return:


 5)                               91.486164
10)                               29.963591

which is 1.91% of what was spent, so for a Gold Award of 50 coins and a jackpot of 100 coins, the return is increased to 74.66%, and the profit decreased to 25.34%. Again, for any other change in the sum of the Gold Award and the jackpot, the change scales linearly.

Incidentally, it might be asked why the choice of a particular sequence of pulls, the one starting from the indicator being at zero, and ending with either a prize being won from the reels, or the prize from the indicator hitting ten being won, produces a percentage that can be meaningfully called "the percentage" of that slot machine.

Of course, this choice can be easily justified. Since the indicator is reset to zero immediately following either the reels awarding a prize, or ten coins being returned because the indicator has reached ten, a very long sequence of consecutive pulls on the machine would, except for short sequences of pulls at the beginning and end, consist of a series of this kind of sequence one after another. Thus, the percentage for that kind of sequence is indeed the limiting value for a sufficiently long sequence of pulls.

This is the sort of thing that is too obvious for words... at least if one is a mathematician.

Another interesting machine from Mills was the Bonus; this had the letters of the word "BONUS" overprinted on five of the symbols on the first reel, and as they appeared, one at a time in order, the word "BONUS" built up in a display at the top. When the word was completed, a special payout of 18 coins was made. An early Mills Bonus slot machine, with the Horsehead style of case, is shown on the left.

This machine remained popular for some time, and thus the BONUS feature continued to be offered in Mills' later machines. The machine illustrated on the right is such a machine, in the "hi-top" style. This style of machine was first introduced in 1947, with the Mills Jewel Bell.

The very first Mills Bonus machines actually used a 10-stop structure similar to that of the Futurity, but in their case, they alternated between the odd and even stops on odd and even plays. This was apparently intended to prevent a lucky player from completing the word BONUS in only five spins. However, a year or two later, either it was recognized that the risk involved of making it appear that these machines were rigged outweighed this issue, or, perhaps even more likely, it was decided that the chance of obtaining the letters B-O-N-U-S in a small number of spins was so low that attempting to avoid that improbable occurence did not justify the extra cost of this mechanism, and so subsequent Mills Bonus slot machines worked in the conventional manner as normal 20-stop machines.

In the case of the conventional Mills Bonus machines, the reels contained the following symbols:

(*) Lemons       4   -   4
(%) Cherries     6   7   -
(O) Oranges      3   5   6
(@) Plums        5   1   6
(A) Bells        1   5   3
(=) Bars         1   2   1

Given the payouts of

%%.         3
%%*, %%A    5
OOO, OO=   10
@@@, @@=   14
AAA, AA=   18
===       100

we can calculate how many coins this paid out for each 8000 coins put in, on average:

Reel           Combinations  Value   Amount
 1    2    3                         Paid

 6 *  7 * 13 =          546  *   3 =  1638
 6 *  7 *  7 =          294  *   5 =  1470
 3 *  5 *  7 =          105  *  10 =  1050
 5 *  1 *  7 =           35  *  14 =   490
 1 *  5 *  4 =           20  *  18 =   360
 1 *  2 *  1 =            2  * 100 =   200

As for the BONUS payments, what proportion of the coins put in do they correspond to?

Assuming a player comes to a machine with no letters showing in the annunciator, and plays, with the normal return of 5208 coins out of every 8000 coins played, how long, on average, will it take for the BONUS combination to complete, returning an extra 18 coins?

Well, since at each step, one particular letter, with one chance of 20 in coming up, is what is needed to advance to the next state, that time is five times as long as it takes for one symbol that appears only once out of 20 possibilities to come up.

Rather than a complicated calculation of the powers of 19/20, however, this suggests that if one treated the BONUS prize as a prize of 18/5 coins, paid whenever the letter B came up, the percentage would be the same.

So that would be 18 coins back from every 100 coins played, or 1440 coins back from every 8000 coins played.

The payouts of the earliest BONUS machines were different:

%%.         3
%%*, %%A   10
OOO, OO=   10
@@@, @@=   14
AAA, AA=   18
===        50

And the symbols on the two halves of each reel were:

                 Symbols together with           Symbols with no
                 the letters B-O-N-U-S           letters

(*) Lemons       -   -   -                       2   -   4
(%) Cherries     4   5   -                       1   3   -
(O) Oranges      1   2   4                       2   1   3
(@) Plums        1   -   5                       3   1   1
(A) Bells        3   2   -                       2   1   1
(=) Bars         1   1   1                       0   4   1

So for every 1000 coins put in when the letters could turn up, one can calculate what is paid out:

         Reel           Combinations  Value   Amount
         1    2    3                          Paid

%%.      4 *  5 * 10 =          200  *   3 =   600
OOO,OO=  1 *  2 *  5 =           10  *  10 =   100
    AA=  3 *  2 *  1 =            6  *  18 =   108
===      1 *  1 *  1 =            1  *  50 =    50

And for every 1000 coins put in when the letters could not turn up, what is paid out is:

         Reel           Combinations  Value   Amount
         1    2    3                          Paid

%%.      1 *  3 *  5 =           15  *   3 =    45
%%A,%%*  1 *  3 *  5 =           15  *  10 =   150
OOO,OO=  2 *  1 *  4 =            8  *  10 =    80
@@@,@@=  3 *  1 *  2 =            6  *  14 =    84
AAA,AA=  2 *  1 *  2 =            4  *  18 =    72

which means the machine paid back 1,281 out of every 2,000 coins, or only 64.05%, not including the BONUS award.

Instead of a letter coming up 1/20 of the time on every spin, it comes up 1/10 of the time on half the spins, so it may still be reasonable to treat its value as 18 coins out of every 100, thus giving the machine an 82% payout.

I have been informed that it was possible to change a cam on this machine, so that instead of the two sets of symbols alternating with every turn, the pattern could be changed, presumably to make the symbols without the letters B-O-N-U-S, and with a lower payout percentage, more likely at the discretion of the operator.

Also, while this special mechanism was used only on the Horsehead style of Bonus Bell, and not on the Hi-Top style, it's important to note that most Horsehead Bonus Bells did not have the special mechanism, so this type of machine is quite rare. Also, before the Hi-Top version of the Mills Bonus Bell came out, but after the end of World War II, Horsehead Bonus machines were introduced in the 50 cent denomination; none of these would have had that mechanism.

It may also be noted that SEGA at one point, much later, made a slot machine that was inspired by the Mills Bonus Bell. It had an electrical display on which the letters B-O-N-U-S lit up one by one; but the display was advanced by an "Advance Bonus" symbol coming up on the first reel, not by the next letter coming up. Of course, either way, there was one position on the reel that advanced the bonus, even if in the original design that position changed at each step, so this in itself would not affect the odds.

As for the Pace Kitty, pictured at right, the symbols on its reels are:

(^) Kitty        4   -   -
(*) Lemons       -   -   4
(%) Cherries     8   7   -
(O) Oranges      2   6   6
(@) Plums        4   2   5
(A) Bells        1   3   3
(=) Bars         1   2   1

The conventional payouts were:

%%.         2
%%*, %%A    4
OOO, OO=    8
@@@, @@=   12
AAA, AA=   16
===        70

assuming the jackpot to be 50 coins, in addition to the regular 20-coin payout for three bars.

Since a Kitty symbol on the first reel caused one coin to be added to the Kitty, which would eventually be returned to the player after 10 losing plays in a row, it can be treated as a combination that rewards one coin to the player. That would, however, lead to an inaccurate calculation of the percentage if the Kitty was likely to overflow before it was paid out. The Kitty had a capacity of 50 coins.

Performing that calculation:

          Reel           Combinations  Value   Amount
          1    2    3                          Paid

^..       4 * 20 * 20           1600  *   1 =  1600
%%.       8 *  7 * 13 =          728  *   2 =  1456
%%*, %%A  8 *  7 *  7 =          392  *   4 =  1568
OOO, OO=  2 *  6 *  7 =           84  *   8 =   672
@@@, @@=  4 *  2 *  6 =           48  *  12 =   490
AAA, AA=  1 *  3 *  4 =           12  *  16 =   192
===       1 *  2 *  1 =            2  *  70 =   140

for a total payout percentage of 76.48%.

Excluding the combination with a Kitty on the first reel, there were 1,266 winning combinatins out of the 8,000 possible combinations. So the chance of ten combinations in a row that did not win would be about 17.8%, which would make it unlikely that, with a 20% chance of adding one coin in every spin, that more than 50 coins would accumulate in the Kitty.

The Modern Era

So far, we have looked only at fully mechanical slot machines, which, of course, are no longer in use.

The sixth set of payouts in the second row of the large illustration near the top of this page are those of the Bally 742A slot machine. They are almost identical to the payouts of the Mills 21 Bell, except that the payout for three melons is larger, and the 7 is now an ordinary symbol occupying a space on the reel in its own right, rather than being overprinted on anything else.

Today's slot machines often have single, double, and triple bar symbols on them, and so I thought I would look for where this started.

The Bally 809, from 1967, was the first machine to allow you to put in from one to five coins for a single spin, multiplying the prizes by the number of coins paid, although an earlier machine from Bally, the Double Up, also known as the Bally 808, allowed one to play one or two coins in this manner.

It was an electromechanical slot machine. That is: the spin of the reels was still the actual randomizing element which decided if a play would win or not. But instead of each reel being associated with a disk with holes, so that three reels with the same symbol would allow a rod to go through three corresponding holes, instead the reels were associated with disks that had notches the depth of which indicated what symbol was printed on the reel strip in each position. The notches allowed a wiper to move and make electrical contact with the one of several contacts that corresponded to the symbol.

This design was first used by Bally in 1946, on the Bally Triple Bell console machine, shown to the right, which meant that, although competing slot-machine makers were still slow to adopt the innovation, it was possible for them to do so when they saw the need as the patents would have expired.

The Bally 809 also had the single, double, and triple bar symbols so common on slot machines today. This machine was one of the earliest to use the double and triple bar symbols, along with the Double Up from 1966. The "Any Bar" symbol used on the actual machine's payout chart to indicate that a single, double, or triple bar may appear in that position is used here, consisting of the words "Any Bar" on two short bars.

The payout percentage of 82.5% was identical on the Bally "Money Honey" slot machine from 1963, which was the first electromechanical slot machine in the conventional format, as opposed to console machines such as the Triple Play. It did not include the multiplier feature, and used a different set of symbols on its reels, not including the double or triple bar.

The Money Honey machine, also known as the Bally 742, is also of interest. One of the symbols was a drawing of a girl's face, the titular "Money Honey". This symbol paid 18 coins if it appeared in any of the three visible positions on all of the three reels, except for paying a jackpot if all three appeared on the pay line; thus, this machine can be thought of as another descendant of the Buckley Criss-Cross. On one version of the machine, a jackpot was also paid for a bar on each of the three reels in any position, but that one had cherries on the third reel.

The symbols on its reels are:

(%) Cherries     2   5   -
(O) Oranges      4   1  10
(@) Plums        5   1   5
(A) Bells        1   8   1
(=) Bars         1   1   1
(*) Stars        2   1   1
(C) Melons       4   2   1
(&) Girls        1   1   1

The machine's payouts were:

%..         2
%%.         5
OOO, OO=   10
@@@, @@=   14
AAA, AA=   18
===       150
***       100
CCC        50
&&&       200

in addition to those for the bars and faces in any position, and so the payout percentage can be calculated as follows:

          Reel           Combinations  Value   Amount
          1    2    3                          Paid

%..       2 * 15 * 20 =          600  *   2 =  1200
%%.       2 *  5 * 20 =          200  *   5 =  1000
OOO, OO=  4 *  1 * 11 =           44  *  10 =   440
@@@, @@=  5 *  1 *  6 =           30  *  14 =   420
AAA, AA=  1 *  8 *  2 =           16  *  18 =   288
===       1 *  1 *  1 =            1  * 150 =   150
***       2 *  1 *  1 =            2  * 100 =   200
CCC       4 *  2 *  1 =            8  *  50 =   400
&&& any   3 *  3 *  3 = 27 - 1 =  26  *  18 =   468
&&&       1 *  1 *  1 =            1  * 200 =   200     

and so it returned 4766 coins out of 8000, for a payout percentage of 59.575%, and thus a profit of 40.425%, however, as there are multiple payout variations of this machine, I may have used the reel strips for a different set of payouts for my calculation.

Thus, for example, if, despite the fact that the example I found called for a cherry to be on the third reel, one changed the payout for three bars to be 100 coins for the three bars in any position, instead of 150 coins for them on the center line, that would add 2650 coins, leading to a payout of 7316 coins from every 8000, still profitable, for a payout percentage of 91.45%. But that is now higher than the 82.5% claimed for the machine.

Having three different kinds of bars, and having a payout of 20 coins (or, for that matter, 18 coins) for any mixture of those three different kinds of bars is, of course, reminiscent of having a payout of 18 or 20 coins for only one kind of bar, but in three different positions.

Of course, there is one important difference: instead of 27 possibilities of which one is a jackpot, now there are 27 possibilities of which three are jackpots.

The 809 had a multiplier feature, where one could wager one to five coins on a single pull of the handle. Thus, it did not have another feature which has since become common on the slot machines today: the 5-line pay progressive feature, where one can pay more than one coin in order to win on additional lines in the window in addition to the center line. This feature also existed on the Bally electro-mechanical slot machines; the Bally 922, from 1971, was one machine with progressive 5-line play. The Bally 831, from 1968, had progressive 3-line play; the first coin won on the conventional pay line, the second on the line above, and the third on the line below. The 922 continued the pattern with the fourth coin paying on the diagonal line going from the top left to the bottom right, and the fifth coin paying on the diagonal from the bottom left to the top right. The jackpot for three sevens was higher for a win on the fifth line, as an inducement to play more coins.

While three bars in any position wouldn't work with that feature, three different kinds of bars instead would be perfectly compatible. Hence, it seems to me that the slot machines of today can trace some of their typical features through to the early Bally electromechanical machines, and beyond them to the fully-mechanical Buckley Criss-Cross.

Although the Bally 922 didn't have the double and triple bar symbols, the Jennings 400 was one example of an electro-mechanical slot machine that combined those symbols with 5-line play. I have seen one source which stated that the Bally Triple Bell allowed 3-line play, but as far as I can tell from looking at images of it, it offered a multiplier feature, and the ability to play using three different denominations of coins, but not 3-line play.

Incidentally, while Bally introuduced the currently popular single, double, and triple bar symbols, the idea of having multiple kinds of bar on a slot machine was introduced at an earlier point. Some Jennings Standard Chief machines used in casinos had both ordinary bars, and special bars - the special bars could both substitute for an ordinary bar, and provide a larger payout if one had three special bars in a row.

The special bars were surmounted by an Indian head of similar form to the one which adorned the cases of earlier pre-war Jennings slot machines from the first Jennings Chief onwards.

The same head design, but in full color instead of metallic grey, without a bar behind it, appeared on the first reel of early Jennings "Wild Indian" machines. Later, a cartoony full-figure image of an agitated brave in buckskin breeches replaced it.

Mills had a Deuce Wild slot machine both in Chrome Bell and Hi-Top models, and there was a Wild Lemon console by Bally, so a symbol on the first reel that could substitute for anything, just as a bar on the third reel could substitute for several things, was another added feature on some slot machines.

In the case of Bally's Wild Lemon console from 1947, however, the lemon only qualified as a wild symbol when the "Wild Lemon" light on the console was lit, not on every pull.

For whatever reason, the 2 on the Mills Deuce Wild machine was overprinted on a bell symbol, instead of standing alone; whether to add visual interest, or for consistency with the 21 Bell (although as that one first appeared as a Hi-Top machine, while there are Chrome Bell Deuce Wild machines, that hardly seems possible, but payout schedules can be retrofitted) or on general principles of some sort is not clear.

The Mills machine even had a payout schedule that explicitly showed the 2 substituting within every possible winning combination. So on the right I show such schedules for all three of these machines, although the other two did not. The actual payout schedule on the Mills machine just used a printed digit 2 rather than showing the symbol as it appeared on the reels - and like most payout schedules on slot machines, it was largely in the reverse of the order I use, starting with the larger prizes and working down to the smaller ones.

Of course, while it hardly needs to be said, as it is well known, it should at least be mentioned, if only for reasons of acknowledgement, that from the 1930s to the 1960s, unlike today, nothing was considered amiss in the commercial use of Native American imagery, as though American Indians, like, say, ancient Egyptian Pharaohs, were figures of the distant history of ancient times... rather than a group of people still in existence, whose own views on the matter ought to be consulted. It was as if the concept of "cultural appropriation", so familiar today, had not even yet been raised... which, I suspect, is not entirely true, even if the phrase was not coined, and the idea was little heeded.

Some Design Examples

Let us suppose that we wished to create a slot machine with a payout schedule like the following:

%..                  2
%%.                  4
%%%, %%*, %%A, %%=   8
OO.                 10
OOO, OO=            20
@@.                 12
@@@, @@=            24
AA.                 14
AAA, AA=            28
===                 50
*A*                 50

The idea is to provide a payout schedule that is very generous, combining the classic and modern approaches to cherries, and also modifying the payouts for the orange, plum, and bell so as to make a three-reel machine behave almost like a four-reel machine.

Can a reasonable distribution of symbols on the reel strips produce a profitable slot machine with this payout schedule?

It isn't easy, but if one takes the extreme measure of putting only one cherry on the first reel, it is barely possible:

If the symbols on the reels are distributed as follows:

(*) Lemons       6   0   1
(%) Cherries     1   6   6
(O) Oranges      5   1   6
(@) Plums        1   6   5 
(A) Bells        6   1   1
(=) Bars         1   6   1

then, the result is:

Reel           Combinations  Value   Amount
 1    2    3                         Paid

 1 * 14 * 20 =          280  *   2 =   560
 1 *  6 * 11 =           66  *   5 =   330
 1 *  6 *  9 =           54  *   8 =   432
 6 *  1 * 13 =           78  *  10 =   780
 6 *  1 *  7 =           42  *  20 =   840
 1 *  6 * 14 =           84  *  12 =  1008
 1 *  6 *  6 =           36  *  24 =   864
 6 *  1 * 18 =          108  *  14 =  1512
 6 *  1 *  2 =           12  *  28 =   336
 1 *  6 *  1 =            6  *  50 =   300
 6 *  1 *  1 =            6  *  50 =   300

giving a house advantage of just over 9.2%.

Of course, making the machine look like a four-reel machine in terms of how it handles the orange, the plum, and the bell, but having to limit jackpots to 50 coins is to omit the most attractive feature of a four-reel slot machine.

However, if one added just one symbol to the machine's repertoire, providing an additional jackpot for three sevens or three melons, that would make it easy to adjust things to allow room for at least that one jackpot to be more generous. Of course, allowing lemons on the first reel and responding to that by making lemon-bell-lemon a jackpot means there is already one extra jackpot after three bars, but then there are real slot machines with both melons and sevens.

Actually, though, it's obvious that only a drastic measure will salvage this design so as to yield something of practical use.

If the payouts for three oranges, three plums, and three bells are restored to their conventional values, then one could still retain payouts for two oranges, two plums, and two bells at half of that, thus still meeting the goal of having winners more frequently than if matching symbols on all three reels are required for nearly every prize, and now have room for a generous jackpot.

As well, the constraint to a maximum of six of any one symbol on any reel would need to be discarded. This could yield something like this:

(*) Lemons       1   0   1
(%) Cherries     1   9   6
(O) Oranges      8   1   6
(@) Plums        1   8   5 
(A) Bells        8   1   1
(=) Bars         1   1   1

Apply to that a payout schedule such as:

%..                  2
%%.                  4
%%%, %%*, %%A, %%=   8
OO.                  5
OOO, OO=            10
@@.                  6
@@@, @@=            12
AA.                  7
AAA, AA=            14
===                100
*A*                200

and the result becomes:

Reel           Combinations  Value   Amount
 1    2    3                         Paid

 1 * 11 * 20 =          220  *   2 =   440
 1 *  9 * 11 =           99  *   5 =   495
 1 *  9 *  9 =           81  *   8 =   648
 8 *  1 * 13 =          104  *   5 =   520
 8 *  1 *  7 =           56  *  10 =   560
 1 *  8 * 14 =          112  *   6 =   672
 1 *  8 *  6 =           48  *  12 =   576
 8 *  1 * 18 =          144  *   7 =  1008
 8 *  1 *  2 =           16  *  14 =   224
 1 *  1 *  1 =            1  * 100 =   100
 1 *  1 *  1 =            1  * 200 =   200

and now the payout percentage is a mere 68.0375%, more like that of a typical slot machine, and allowing some room to adjust things upward. But not enough room to allow even so much as a second cherry on the first reel, so it's still a tight squeeze.

Now, let's look at another possibility. This one is more complicated, so it may not be possible to make a purely mechanical slot machine of ordinary size with such a payout schedule. Perhaps it would instead require at the least an electromechanical console.

Here, the motivation is this:

Unfortunately, despite being creative and ground-breaking, the AC Novelty Seven-Way Multi-Bell did not prove to be a great success in the marketplace, and I presume this is because it was too different and unfamiliar.

The Mills 4 Bells, on the other hand, also let the player choose, to a limited extent, whether to bet on oranges, plums, bells, or bars.

So in a way, it was similar to the Seven-Way Multi-Bell. However, picking a particular symbol only doubled (or, in one case, tripled) the regular prize for that symbol, so it really wasn't that different from a conventional slot machine.

So the idea behind the payout schedule shown at right is to offer the player five choices, one for every symbol that appears on the reels, and, while largely retaining the conventional payout structure of an ordinary slot machine, to still move to some extent in the direction of the Seven-Way Multi-Bell in that when the player chooses one of the five symbols on which to bet, a prize, although small, will be won even if that symbol only comes up on the first reel.

Aside from questions of implementation, however, there is another reason to doubt whether such a payout schedule is even possible.

Except for the very earliest payout schedules, where the prizes were relatively small, and there were no jackpots, it was necessary to decrease how often winning combinations came up by skewing the distribution of symbols on the wheels; so instead of having, say, five oranges and four plums on each of the first two reels, one might have eight oranges and one plum on the first reel, and two oranges and seven plums on the second reel.

But for a payout schedule like the one shown at right to make sense, it would seem that a relatively even distribution of symbols on the reels is required, making it impossible for a profitable arrangement to be constructed. And, indeed, I tried that, and it definitely did not work. Perhaps, though, the distribution of symbols could be made uneven in such a way that the additional payout for each chosen symbol is equal.

Let's see what would happen for one possible arrangement:

(%) Cherries    3  5  6
(O) Oranges     4  4  4
(@) Plums       4  4  3
(A) Bells       3  6  6
(=) Bars        6  1  1

For the basic payouts, clearly this distribution will be highly profitable:

           Reel           Combinations  Value   Amount
            1    2    3                         Paid

%%.         3 *  3 * 14 =          126 *    2 =    252
%%%         3 *  3 *  6 =           54 *    4 =    216
OOO, OO=    6 *  5 *  4 =          120 *   10 =   1200
@@@, @@=    3 *  5 *  5 =           75 *   14 =   1050
AAA, AA=    2 *  6 *  7 =           84 *   18 =   1512
===         6 *  1 *  1 =            6 *   50 =    300

The remaining number of coins is 3470, so that is the limit for any of the additional payouts.

           Reel           Combinations  Value   Amount
            1    2    3                         Paid

%..         3 * 17 * 20 =         1020 *    3 =   3060
%%.         3 *  3 * 14 =          126 *   +3 =    378
%%%         3 *  3 *  6 =           54 *   +4 =    216

O..         6 * 15 * 20 =         1800 *    2 =   3600
OO.         6 *  5 * 15 =          480 *    5 =   2400
OOO, OO=    6 *  5 *  5 =          120 *  +10 =   1200

@..         4 * 15 * 20 =         1200 *    2 =   2400
@@.         4 *  5 * 15 =          300 *    7 =   2100
@@@, @@=    4 *  5 *  5 =          100 *  +14 =   1400

A..         4 * 16 * 20 =         1280 *    2 =   2560
AA.         4 *  4 * 13 =          208 *    9 =   1872
AAA, AA=    4 *  4 *  7 =          112 *  +18 =   2016

=..         1 * 19 * 20 =          380 *    3 =   1140
==.         1 *  1 * 19 =           19 *   10 =    190
===         1 *  1 *  1 =            1 * +150 =    150

If we change the payout in the case where cherries are chosen to 2 coins instead of 3 coins for a single cherry, we lower the additional payout by 1020 coins, from 3654 coins to 2634 coins, which allows that choice to be profitable.

[Next] [Up/Previous] [Mathematics] [Other] [Home]