[Home]

Featured Images

On another page, I discuss digital magnetic tape recording. Also, in discussing 5-level code and ASCII, my diagrams include the paper tape formats associated with these codes.

However, there isn't quite enough to say about paper tape formats to fill a page... although this particular paper tape format might be complicated enough.

However, I'm not quite enough of an antique collector to be an expert on these. However, hunting around and finding different sources on the Web enabled me to compile the information for the diagram above.

One of the biggest companies in the player piano field in the United States was the Aeolian company, which trademarked the term Pianola, and which later made the Duo-Art line of player pianos.

In December of 1908, a meeting between the major companies that made player pianos took place at the Iroquois Hotel in Buffalo, N.Y., which led to an agreement called the "Buffalo Convention", standardizing the format of piano rolls for player pianos with a full gamut of 88 notes, to use a hole spacing of 9 holes to the inch, and to use the same 11 1/4 inch (or, from one source, 11 9/32 inch) width of roll as used for existing Aeolian piano player rolls with a 65 note gamut and a hole spacing of 6 holes per inch, which had been to some extent the de facto standard of the industry, although there had been many incompatible formats of roll for player pianos in use up to that time.

There was controversy over the choice of a 9 hole per inch spacing. While the older piano rolls, with a spacing of six holes per inch, could be kept in alignment through normal mechanical design of the piano, a nine hole per inch spacing (but not necessarily an eight hole per inch spacing) would make it necessary for the piano to dynamically align the piano roll through some type of self-correcting mechanism.

Aeolian player pianos had a trademarked feature called the "Themodist" where one position on each side of the player piano roll (usually activated by a distinctive pattern of two small holes instead of one hole) indicated that any notes struck at the same time as indicated by that hole would be emphasized. Each position governed the notes on one side of the piano, so that specific notes in the main melody or theme, and specific notes in the chords or accompaniment, could be independently emphasized.

Another relatively common player piano feature was a pointer that the operator of the piano could slide across the roll to follow a line printed on it which indicated how the tempo of the music should vary during performance.

The Aeolian Duo-Art player piano, in addition, also allowed the user to either play normal 88-note piano rolls, or, by switching over some pneumatic connections, to reserve the positions corresponding to the bottom four notes, and the top four notes, of the piano for use to indicate how forcefully the notes of the theme and accompaniment would be struck in general. 16 levels were available for both.

The position on the left-hand side shown as used for the sustain pedal appears to have been used on other types of player piano as well which did not have other reproducing piano features. Also note the position on the right-hand side for the soft pedal.

The diagram shows how the notes on an 88-key piano correspond to the hole positions on a player piano roll, including how the 65-note gamut fits into the 88-key gamut of a piano, and how a modern 61-key keyboard fits within that 65-note gamut. Yellow and blue keys are those on the 88-key keyboard and not the 65-key keyboard; cyan and purple keys are those on the 65-key keyboard and not the 61-key keyboard. Middle C is indicated by making that key green.

The top part of the diagram shows the expression system offered by Ampico. There were three holes on each side used to determine the force with which notes were struck; they were labelled 2, 4, and 6 based on their hole positions. They also added in a binary fashion, having the values 1, 2, and 4, but the result of the addition then related to the force with which the keys were struck according to one of three response curves, which could be selected by a lever, so that the instructions given with a particular piano roll could specify either a larger range of dynamics, or a finer control over dynamics, would be used with that roll.

These holes caused the level to latch; another hole, considered to be in position 7, cleared the level. As well, there was another expression control, crescendo, and the speed with which it went on or off was controlled by the hole marked fast. The original Ampico design had independent crescendo functions for both sides of the piano, but this was later replaced by having only one crescendo.

Note that the use of hole 3 on the left for the sustain pedal put this in about the same position as used on Duo-Art pianos.

Working Ampico reproducing pianos are scarcer than Duo-Art reproducing pianos, as, according to a web site I have seen, it is more difficult to restore them well enough to produce acceptable results, although originally both kinds of player piano were fully comparable.




In Canada, for an extensive period of time, the game Monopoly (a trademark of Hasbro; then, of course, of Parker Brothers) was sold with relatively conventional wooden game pieces, rather than metal tokens. As this set of pieces is relatively obscure outside Canada (although for brief periods wooden pieces, some of which were similar, were used in the U.S.) I have pictured it here.

I believe this set of pieces was also used with sets for the game Clue, also a trademark of Hasbro at this time.

The above image has been split, with a part in black and white, to prevent fraudulent use in online auctions.

A little mathematical artwork I've come up with; I call it The Electric Dodecahedron.

I was a bit disappointed that no Archimedian solid includes both octagons and pentagons. And so, after some thinking, I came up with this shape:

Start with the 30-sided rhombic triacontahedron, and put an octagon (shown in green) on each of its diamond-shaped faces. Where three obtuse angles of the diamonds meet, an equilateral triangle (shown in yellow) is formed. Then, by suitably joining other corners of the octagons, a decagon (shown in red) with alternating long and short sides, still having pentagonal symmetry, is formed. The shape is less elegant than an Archimedian solid, as there are left-over isosceles triangles (shown in white) with narrow bases present as well. There are 30 octagons, 20 equilateral triangles, 12 decagons with alternating sides, and 60 narrow isosceles triangles in this shape. Since only the octagons and decagons are large shapes, this could form the basis of a somewhat imperfect 42-sided die, having pentagons and bilaterally-symmetric, but not regular, hexagons as faces. Perhaps if they ever make the Hitchhiker's Guide to the Galaxy into an RPG, they might use such a shape. Having seen this interesting site, however, I thought perhaps it might have already been done, but it seems not to have been.

The shape depicted, however, is indeed one that has been previously known; it is designated as I(*,1,4,e) in a paper on "symmetrohedra" by Craig S. Kaplan and George W. Hart.

Speaking of useful shapes for dice, since 7 times 20 is 140, and 5 times 12 is 60, and 140 plus 60 is 200, one could start from the shape of a soccer ball, following this principle:

and obtain another imperfect die, but one that it seems could be made quite close to a regular design, with 200 faces.

Here's an image of the graticule of a very clever circular slide rule for vector addition. Basically, take polar-coordinate graph paper, and bend it around into a circle... this is the Jensen aircraft computer.

This image shows how five sets of three great circles can be disposed symmetrically about the sphere, using the "soccer ball" Archimedian polyhedron as the base.

To the right is a picture of the cylinder of a steam engine. A valve lets steam in on one side of the cylinder, and lets spent steam out on the other side of the cylinder. Since the piston moves from one end of the cylinder to the other as it is pushed from one side, and then back again pushed from the other, the motion of the valve has to lead that of the piston by 90 degrees in phase.

If you remember pictures of steam turbines, you will recall that there are many layers of vanes, each one larger than the one before. This is to continue extracting energy from the incoming steam as it gradually loses pressure. Reciprocating steam engines, particularly when used on ships, could do the same thing by having three cylinders, in each of which the steam lost part of its heat and pressure, before the spent steam was pulled from the last one by the condenser. The picture at the right illustrates such an arrangement. The smaller the change in the pressure and temperature of the steam in each cylinder, the closer the cycle approaches reversibility, which is the condition for the maximum theoretical efficiency, that of the Carnot cycle.

All the cylinders must operate in the same phase, simply because as the wheel turns, the piston is at a position, and moves at a speed, proportional to a trigonometric function of the phase. The cylinders are illustrated at phases spaced by 120 degrees to show more of the operation of the engine.

But the real secret of how a steam engine works is not visible anywhere on the diagram at right, and usually is not mentioned in most short descriptions of how a steam engine works!

Steam engines on steam locomotives just let the spent steam out of a smokestack as they went along, and frequently obtained new supplies of water. But most steam engines recirculated the water that served as their working fluid.

If you have a hot boiler, out of which steam is coming at a high pressure, then that pressure is the same in all directions everywhere in the boiler! So how do you push the recirculated water back into the boiler?

The short answer is, with a pump. But the next question that might come to mind is how you avoid, then, using up all the energy your steam engine produced in trying to pump the water back in to where it came from.

And the secret of how a steam engine works is this: steam made from water takes up about 2,000 times as much volume as the water it came from. Since the amount of work required to cram something into an area at high pressure is proportional to its volume, because steam takes up so much more volume than water, pumping the water into the boiler only uses up a tiny amount of work compared to what the steam engine produces, even if the steam engine is only 4% efficient in the first place, as was true of the first modern steam engines as invented by James Watt! As the increasing size of the cylinders indicate, though, unlike water, steam is compressible; but even one part in 500 instead of one part in 2,000 is still inconsiderable enough.


A special kind of paper tape, having room for only two holes across, was used for recording Morse code. However, this tape came in two very different forms, which has led to some confusion.

On the top, the Wheatstone form of this tape is shown. This tape used a hole on the top to represent the start of a dot or dash, and a hole on the bottom to represent the end of a dot or dash. This kind of tape, moving at a uniform speed through an automatic sender, would produce dots and dashes in the ideal 1:3 ratio.

On the bottom, the form of tape used for sending messages in cable code is shown. Here, a dot is represented by a current in one direction, and a dash by a current in the opposite direction, or down another wire, each taking an equal time. In a sense, cable code isn't "really" Morse code, but it used the existing International Morse Code as its basis, rather than inventing a new code. When in use, this tape also moved at uniform speed through transmitting equipment, which made such equipment incompatible with any attempt to use it to transmit conventional Morse code.

Both tapes are shown with the name "Morse" encoded on them. Of course, Morse himself only invented the American Morse Code, not the International Morse Code; it might be possible to coax the latter from a Wheatstone device, but a code with different lengths of dashes is quite incompatible with cable code.


From the Book of Ezekiel, chapter I, verses 16 and 17:

The appearance of the wheels and their work was like unto the color of a beryl: and they four had one likeness: and their appearance and their work was as it were a wheel in the middle of a wheel.

When they went, they went upon their four sides: and they turned not when they went.

In a mechanical computer mouse, the ball that moves when the mouse is moved on a surface is within wheels that sense its vertical and horizontal movement; a mouse, unlike a toy truck, is not rotated when one changes from moving it up and down to moving it from side to side. Could the Prophet Ezekiel have been describing something based on a similar principle?

Of course, a normal wheel, rather than a ball, that can itself turn, being gimballed within a wheel, is another possibility; that which the wheel bears does not turn, but the wheel itself can.

Another mystery from the Bible concerns the 144,000 sealed, 12,000 from each of the tribes of Israel, who are standing in a square. While 144 is a square number, and so are 100 and 10,000, 1,000 is not a square. So how can 144,000 people be standing in a square array?

Well, they don't need to be standing in a rigid grid formation. Thus, the illustration above shows a checkerboard pattern, in which a grid is formed from small, square-shaped clusters of ten people. 14,400 such clusters would make a square with 120 such clusters on each side.

Also, when people are standing up, they're usually wider than they are thick:

Admittedly, though, a 5:2 ratio is perhaps a bit on the large size.

But then, that could be where the factor of 144 comes to our rescue!

144 is 12 times 12. But it's also 4 times 3 times 4 times 3... so it's 16 times 9.

And 16 is not equal to 9, it's bigger than 9. So let's take that, and mix it together with the 5 to 2 ratio so as to moderate it... 16 times 2 is 32; 9 times 5 is 45.

Thus, one can have a square in which each person is placed in a rectangle giving 45 units of width and 32 units of depth - and they stand in 45 files in 32 ranks, forming a square.

Each such square would contain 1,440 people, and so a hundred such squares, in a 10 by 10 array, would be a square with 144,000 people.

Here are 16 files, and 24 ranks along those files, with a dotted diagonal line showing that 16 files wide is the same as 22 and a half ranks deep.

While on the topic of refuting the Biblical naysayers, it might be noted that while a molten sea, ten cubits from brim to brim, would indeed have a circumference of 31.14159... cubits, a molten sea 9.5 cubits from brim to brim would have a circumference of 29.84513... cubits; thus, a molten sea with a diameter anywhere between 9.5 cubits and 9.70845... cubits would have a diameter from 29.84513... cubits to 30.5 cubits, and thus it is entirely possible for a molten sea ten cubits from brim to brim to be compassed round about by thirty cubits if one is rounding to whole cubits.

In ancient Egypt, at least, there were six spans (about 3") to a cubit (about 18"), and there were four digits (about 3/4") to a span. As the size of human hands has not changed much since ancient Egypt, 3/4", or 19.05 millimetres, is also the normal spacing between keys on a typewriter (or computer) keyboard.

Using these units, here is a table covering the range of interest.

Diameter                 Circumference
cubits spans digits      cubits spans digits
  9      3                29      5     0.283125
  9      3     1          29      5     3.424718
  9      3     2          30      0     2.566310
  9      3     3          30      1     1.707903
  9      4                30      2     0.849496
  9      4     1          30      2     3.991088   

Note that a molten sea nine cubits, three spans, and five digits across is compassed round about by less than a hundredth part of a digit less than thirty cubits and three spans. Thus, pi is close to 732/233.


And (at one time) finally, here is

the message encoded in the stones of the Great Pyramid: "The Pharaoh is cheap and wants a 15% discount whenever he buys cloth".

In the novel The Library of Babel, author Jorge Luis Borges describes a library filled with every possible book, in which members of a religious order search for a book which makes sense.

The library is described as being made of hexagonal rooms, with a circular opening in the floor, surrounded by a railing, with a corresponding hole in the ceiling, serving for ventilation.

Each room is said to have bookshelves on four sides, and an opening into a vestibule which proceeds to another room with books on another side; this vestibule has a bedroom and a washroom on either side of it. In the middle, there is a spiral staircase by which the other floors of the library might be reached. (Some translations vary this description from that which I have paraphrased here.)

If we assume the remaining side, not described, is the simplest thing, a doorway directly into another one of the rooms with shelves, it might be that the plan of the library is something like this:

In this way, except for whether the bedroom is counterclockwise or clockwise from the washroom, each of the rooms with bookshelves in the library could be identical to each other room, as long as the direction to the vestibule rotates uniformly from one floor to the next, and rooms with bookshelves can be only above and below other rooms with bookshelves, while vestibules can be only above and below other vestibules.

However, I received an E-mail from David Shalcross, who I thank for helpfully pointing out that there is a flaw in the design above; while motion in all directions is possible, one can only get to every third row of hexagonal chambers on any level through the spiral staircases.

Fortunately, there is a way to correct this simply enough:

If the door between library chambers is on a wall adjacent to the door to the vestibule, then the vestibules are still connected in rows, but now in rows turned 60 degrees clockwise from the rows as they were in the diagram above.

It is not even necessary to use all six kinds of floor; the new rows, since they proceed in a zig-zag path, intersect each and every row, not every third row, of the rows in a different direction on the other floors. (In fact, only two kinds of floors are required to access every room, but all three are required for making the rooms symmetrical.) The old floor style, however, might be useful to facilitate faster horizontal movement.

In each of the three component diagrams in the illustrations above, one of the paths available within each floor is highlighted in a light blue-green color. This change from the initial form of the diagrams helps make the flaw in the first arrangement easier to see.


In the illustration below, first you see a line with a ? as the prompt, and then a BASIC statement with line number entered by the user.

This shows how a simple character-oriented display on a terminal works, with the screen divided up into a grid of characters, all of equal width, as would be made by an ordinary typewriter (as opposed, say, to a proportional-spacing IBM Executive typewriter).

Many terminals supported the APL language, but in general, those video terminals (as opposed to printing terminals, which could simply overstrike on the paper) that did so were designed by one computer manufacturer for its own computers, and had special characters representing the overstrikes that its dialect of APL used.

The next group of lines shows, however, that if a display had two pages of text memory, and a suitable character set (presumably, two copies of the ROM containing it would be needed), then the use of an OR gate could allow arbitrary overstrikes. The idea had been thought of, but there was not enough demand, apparently, for any terminal manufacturer to make such a terminal.

The third group of lines shows that I feel they were not imaginative enough. If one has two pages of text memory, were one to have a somewhat extravagant character set (including superscript characters, and italic characters in both normal and superscript forms) and overstrikes at an offset of one-half of the height of a line (a characteristic, in fact, of most typewriters, so that one could select 1 1/2 line spacing), then one could actually display, without going to a full graphical display as is used with today's graphical user interfaces, mathematical formulas in all their glory, as I show with the definition of a Bessel function, which allows me to show both a sum and an integral.

In the example shown, it is assumed that there are only superscript characters, and subscripts are achieved by using those characters in the row half a line down. This would conflict with the fraction bar when superscripts are used in the numerator expression, so that is not really a reasonable assumption.

Some character (rather than graphic) terminals, such as the HP 2645, did provide a special character set for displaying equations, but they did so with conventional character spacing.


This was inspired by the 4-line Mathematics feature which Monotype provided for its typesetting equipment.

Mathematical notation was difficult to set in movable type. The 4-line mathematics feature did not eliminate all the difficulties, but it did simplify several parts of the process.

A Monotype caster produces individual pieces of type as called for by a punched tape produced by a Monotype keyboard. This means that, unlike Linotype, while it allowed type to be set largely just by typing the text to be set, it also provided the same flexibility, if required, as could be obtained by setting type by hand.

The most dramatic benefit of the Monotype 4-line mathematics feature was that it allowed a variable to have both a superscript and a subscript at the same time. This was achieved as follows:

A line of mathematical notation was allowed 12 points of space. Normal characters were in 10-point type, superscripts and subscripts in characters the size of those of 5 1/2-point type, and second-order superscripts and subscripts were the size of 3-point type.

Characters were, however, cast on a 6-point body, not a 12-point body.

This meant that that the typesetter needed to type the line of mathematics twice, holding down a shift key to blank out characters which belonged to the other 6-point half of the 12-point line. It also meant that when regular-sized characters were cast, part of the character had empty space below it instead of a type slug; this, called "kerning", takes place in the horizontal direction instead of the vertical direction for letters like an italic lowercase f in normal typesetting.

This, in itself, was not an unprecedented feature. Even the Linotype machine allowed one to place double-height characters on single-height bodies; this was done to allow prices to be more prominent in advertisements otherwise set in small type, and so the high characters were called "advertising figures". Because of this, it was possible to typeset simple mathematical formulas, such as might be needed in high school textbooks, on a Linotype. But the Monotype, with much greater flexibility in spacing, was the only choice for publishing advanced textbooks or scientific journals which required essentially an unrestricted ability to typeset mathematics.

Placing larger characters on smaller bodies allowed one of the two typings of a line to include superscripts, and the second typing to include subscripts. It also meant that only two copies of a character were needed, rather than four, to allow superscripts and subscripts to have their own superscripts and subscripts - which could still be typed directly in the line, instead of being added later.

In addition to the direct benefits of the technique, Monotype 4-line mathematics was significant because it involved the creation of new fonts of characters for mathematical composition in Times Roman. Previously, mathematical typesetting on the Monotype tended to be done in Modern No. 7.

A new set of italics was designed for the Monotype Times 569 as distinct from the regular Times Roman font, Monotype Times 327. The lower-case italic v was changed so that it would be more distinct from the Greek letter nu, w was changed to match it, and g was changed to have the shape used in hand printing rather than the double-loop shape used in many printed fonts. As well, all the italics were redesigned to be less slanted, to reduce cases where extra spaces would need to be inserted due to horizontal kerns.


Just in time for Hallowe'en: a trapezohedron whose faces are not trapeziums, but trapezoids - even if somewhat unsymmetrical ones:

Using only symmetrical trapezoids, one can construct a distorted version of a cube or rectangular prism with two identical faces and two parallel faces, and one can also join two such shapes by their largest face to form a 10-sided solid whose faces are symmetrical trapezoids. The diagram below shows the view of such a shape when looking directly at one of the two parallel faces:

Note how the two tilted trapezoids on the sides, although symmetrical, do not appear to be; thus, on the right side of the diagram, a geometric construction is shown as an aid in overcoming the optical illusion.

Actually, quite a bit of freedom is available in designing such a shape... allowing, for example, an 18-sided solid, or a 14-sided one as shown below.

And, incidentally, the base plan is the plan from the diagram above, just turned upside down (and with a slightly different width), but in this form, the optical illusion of asymmetry is even stronger.

However, from the story The Haunter of the Dark, it is noted that the Shining Trapezohedron is at least roughly spherical in shape, so a clearly rectangular shape such as this is not a possibility.

And while I'm on the topic of H. P. Lovecraft: the state of Massachusetts is justly proud of being the home of Harvard University, in the city of Cambridge, Massachusetts. Harvard's rival, Yale, is located in New Haven, Connecticut. In many of his horror stories, H. P. Lovecraft, an author from Providence, in Rhode Island, adjacent to both Massachusetts and Connecticut, made reference to a ficticious institution of higher learning called Miskatonic University. The Housatonic river is the only river in the area with a similar name, and that river flows through the south-western corner of Massachussetts, and continues through Connecticut to the sea... its mouth being beside New Haven. However, Miskatonic University is located in the ficticious city of Arkham, Massachusetts, said to be modelled on Lovecraft's native Providence, Rhode Island by some, and on Salem, Massachusetts by others. In The Dunwich Horror, both Miskatonic University and Harvard (in the Widener Library) are said to posess copies of the Necronomicon. As it happens, Providence, Rhode Island is home to another Ivy League college, Brown University.

As for the other members of the Ivy League, Dartmouth College is in Hanover, New Hampshire; Columbia University is in New York City, Cornell University is in Ithaca, New York, Princeton University is in Princeton, New Jersey, and the University of Pennsylvania is in Philadelphia, Pennsylvania.


And here is an image of a very common object, generally taken completely for granted, but with some measurements added, measurements that were not easy for me to find:

The outlet on the top is the normal one in use in North America, except that I have illustrated the 20 ampere variant instead of the normal 15 ampere one, by making the opening for a vertical prong on the left have a T-shape.

Some older outlets, without an opening for a ground prong, give a similar T shape to both prongs. The reason they did that was because the oldest common form of an electrical plug had the two prongs horizontal. Today, this form of plug is only used for 250 volt outlets, not for outlets with a normal 120/110/100 volt RMS AC supply, and so that form of outlet is no longer made.

The outlet on the bottom is a type currently in use in Australia, Argentina, China and a few other places for 240 volt 50 Hz AC. As this type of outlet originated in the United States as an early form of grounded outlet, before the round pin was added to the existing plug, it was originally used with the same type of faceplate as the conventional North American outlet. Today, this outlet type is usually used with different styles of faceplate because the electrical codes in the countries which use it mandate additional safety features; I show it here because it is the one outlet for European style AC which doesn't necessarily require a completely different form of faceplate than that used in North America.

If I'm going to provide an image of an electrical plug, though, I should also provide an image of something a little more recondite:

Here are some of the layouts of the pins on the bases of vacuum tubes. In the top row, that of the classic octal tube is given, followed by 7-pin and 9-pin miniature tubes, and then the Compactron tube from General Electric.

In the bottom row, the UX4 and UX6 pin layouts which are two of those which preceded the octal base, found on some American antique radios, are shown.


Copyright (c) 2005, 2008, 2014 John J. G. Savard