circa 1912 - please respect my Copyright.  Thank you.

The Teletyper or Mechanical Cypherer was the very first crypto machine trialled in the Royal Navy.  Automatically codes and decodes by a simple transposition of letters. 

The word 'teletyper' was chosen so that each of the several manufacturers of the kit wouldn't know what the part they were manufacturing was really for, the machine being the sum of its parts. 12 sets were ordered so that sea trials could be carried out. 

How does it work ? - a help with the explanation!

Two ordinary Service type typewriters are connected together side by side [Left and Right] which includes their respective rollers connected as one roller controlled by the Left typewriter which is called the PRIMARY.
Over the top of the Primary are placed 26 spring loaded Plungers a la  

      Note the grey coloured key cover on top of the device which is the same as an ordinary typewriter keyboard key which you press down on. This action concertina's the spring and the little yellow button on the very bottom extends and depresses the typewriter key immediately below it. Here we show just five of the 26 spring loaded Plungers which are called the PRIMARY SWITCHES connecting A with A, U with U, L with L, B with B and C with C, and for ease of drawing only, the 26 keys of the alphabet {each with a second function depending upon the shift for upper or lower casing} are shown on each of the two keyboards shown, 10 on top row {or should be but I had no room for the Letter 'J'}, 9 on middle row and 7 on bottom row. 

Thus, one can read the full alphabet on each keyboard, reading from left to right A-J on top line; K-S on middle line and T-Z on bottom line.

The diagram shows the PRIMARY KEYBOARD sitting above the left hand TYPEWRITERS KEYBOARD so that if the letter 'A' is pressed on the primary it MECHANICALLY presses the letter 'A' on the typewriter keyboard immediately below it and PRINTS the letter 'A' on the left hand roller. It also sends an ELECTRICAL SIGNAL to the Magic Box which we have shown {in a most exaggerated manner} as a simple switch and the primary key NOT yet pressed; when it is, that switch closes and completes the circuit.. The right hand TYPEWRITERS KEYBOARD prints ELECTRICALLY on the right hand roller [which is directly connected to the primary roller] the transposed letter of the alphabet [which represent 'A'] dictated by what is the Magic Box above - say, the letter 'Q'.

We haven't got to the right hand typewriter yet, so if you have followed that we are both doing well !

Over the keyboard of the right hand typewriter there is a group of hands-free electrically operated switches {they too are plungers] each key having its own - so again, 26 in number. Each one is connected to that Magic box above. If the magic box CODES the letter 'A' as the letter 'Q' [as in our example] the Magic Box will send an electrical signal to the letter 'Q' switch which will strike the typewriter keyboard letter 'Q' to PRINT the letter 'Q' on the right hand roller. To DECODE the operator types the received code into the primary; there is a mechanical action on its roller, the electrical signal is sent to the Magic Box and the Magic Box sends the transposed decoded letter to the right hand typewriter, and it, via electricity, prints the plain language on the right hand roller.

By now, those of you with TYPE X/CCM or KL7 experience will be shouting out easy-peasy, let's have more !  For the rest of you it is easy going eh?

Now, all we have to do is to understand what the Magic Box does and then we have cracked it: pardon the pun.
I am wearing my BLUE PETER hat so you know what that means ? First we are going to learn how their [1912] key card worked which they called a CODE CARD or a CARD SWITCHBOARD.
First take two sheets of A4 paper, marry them together, measure 8¼" from the top, draw a line and cut off the smaller remaining sections thereby make two squares. Leave one of the squares blank. On the other square, pierce three good sized holes at random anywhere on the square. Turn the square over and cut or tear off the pushed-through torn paper. On one side mark it TOP then North South East and West, turn over and mark the other side BOTTOM and again N,S,E and W. Place the marked square over the unmarked square so that the word TOP is uppermost and North to the top. Using a pen or pencil mark through the holes you have made leaving a mark on the clean bottom square. Turn the top square through 90˚and again mark the bottom square through the three holes. Continue turning through 90˚, then turn over so that the word BOTTOM is uppermost and repeat through 90˚.  From those simple three holes you have made a generous pattern on the bottom sheet. Now imagine that you had made 26 holes.  That one square with 26 holes has eight days of different code depending upon the orientation of the card.  Thus four such sheets of dissimilar holes would cover a month of crypto.

Ok, but how do we use it?

Imagine 26 wires coloured blue laying on a flat surface each uniformly separated. Above them but not touching  are another set of 26 wires this time coloured red laying at right angles.
The wires can be made to touch each other creating a permutation of 26 x 26 = 676 choices of connections at their cross-over points. If we were to put a kind of switch for each possible junction we could switch on or off at will for that cross-over point. This they did by using a spring loaded plunger and when the card [which we have just made in our Blue Peter session] was not in place all the plungers were proud and sticking up. An analogy here is an openly running autohead  RATT or CW where ALL the peckers are un-hindered and are allowed to cycle without a paper tape forcing them down. Once the card was placed in the switchboard ONLY 26 of the spring loaded plungers were allowed through the holes in the card and a plunger proud was a switch switched on. These switches sent the necessary transposed letter to the electrically operated keyboard on the right hand typewriter printing that letter on the right hand roller. To reverse the processes, the Magic Box had a change over switch marked CODE and DECODE. 

Today it is a "noddy machine" but to men of those far away days it was state of the art stuff.  However, comments were made as follows. It was thought that with such a simple code the system would be vulnerable to crypto-analysis and thus objections were raised.  The objections were not considered to be serious as applied to this apparatus because of the changing code card and the unlimited codes available.  In practice it could be arranged for the code card to be changed at stated times and frequent intervals, or, as an alternative, the number of the code card used could be quoted at the beginning of every message.  Owing to the large number of different combinations available it would be possible to issue complete sets of new code cards from time to time especially if it was feared that any of the cards had been stolen.  {Clearly we adopted their ideas and procedures}.  This didn't please all and HMS Vernon designed a system of changing the code after every letter had been transposed.  That idea was abandoned as unreliable and the code card theory prevailed and the Navy's first stab at a crypto machine joined the fleet. In the end there was only one problem caused by such a machine and that was that it produced a lot of code which kept the ether extremely busy, whereas, using the ordinary Service three letter coded groups direct from books was much quicker - though obviously a lot less secure.  Being a technical man of the W/T Branch, I reckon that those ancient signallers, the Buntings, still back in the 19th century, had a hand to play in its demise !