Adsumudi
This is the machine that belongs to the Arithmeum, serial n° 1636, and that had gotten a bit stiff and sticky 22 years after the last cleaning and restoration. I cleaned it, and will be demonstrating its use in a video at the bottom of this page.The Adsumudi was conceived by Alois Salcher in 1905/1906, and manufactured in 1907. There are some patents relating to the machine from before its manufacture in 1906 (DE204333), as it was produced e.g. GB190623173, FR370829, AT35115 as well as some patents for an improved machine, in which the entire principle of operation has been changed, and that was apparently never manufactured, GB190906657. Various sources mention that the machine was never much of a success, and that that is the reason why it is so rare today. The known serial numbers seem to indicate a production run of about 3-400 machines, and despite production being stopped since 1908, they were still advertised and sold by the end of 1910 (see the Rechnerlexikon).
The machine has a operating principle that is entirely different from any other calculating machine ever manufactured. In order to move the gears with the result wheels attached to them, it has rectangular plates with a slot in the middle, and a gear rack on either side of the slot. This is a picture from deep inside the machine that shows this operating principle:
Depending on which way the result register moves, the gear engages with the rack on the left or the right of the slot. This reverses the direction of the result register from addition to subtraction and vice versa. The racks are connected to the setting levers, which are spring-loaded, so as soon as they are released, they rezero themselves and transfer their value to the result. All the rest of the complicated mechanism is designed to engage and disengage the correct side of the rack with the result at the correct time, and to make sure the result register is locked when it is not in enagement with the racks.
At first sight, there is no way on the outside of the machine to switch it from addition to subtraction, but reading the patent will reveal all - the trick is in a slight rotation of the actuating lever. You can see it has a brass attachement with an engraved S and A, which will alternately show up underneath a window in the top plate when the operating handle is clicked from left to right and back.
Underneath, this is what is happening in the mean time:
The picture above has the operating lever in the rest position. Notice the steel pin, which is sitting on the left with the lever on "A". Below, the operating lever is rotated to S, and partially pushed upwards (down in the picture). You can see that the pin buts up to the slope on top of the steel fork and pushes it (and the result register with it) to the left.
With the operating lever to "A" however, the pin arrives at the other side of the fork and pushes it (and the result register) to the right, for addition.
The machine is arranged with the operating handle at the bottom left, the input pins at the bottom with the control register right above it, and then there are two more registers, the result at the top, and then in the middle a memory register. This can be used to transfer (add) the contents of the result into when clearing the result register. Five knobs are sticking through the wooden case. On the left, there are three. A quick release for the input at the bottom, in the middle a button labeled "Umschaltung" for engaging the memory with the result, and at the top a clearing button for the memory. On the left side there is the zeroing button for the result at the top, as well as a button that allows to disengage the ratchet on the operating handle should something block and it is necessary to reverse the machine back to its rest position. This is not normally possible due to a safety device that can be disengaged with this knob.
With all of this provided, the machine can work as a two-function calculator, eminently suitable for addition and subtraction. Some people seem to be of the opinion that the machine would have been sold like this as a simple adder. The name however, indicates that the machine should also be able to do multiplication and division, and for this an extra tool is needed, namely a template for repeating the setting of numbers in the input. This takes the form of a sliding carriage, which sits on vertical rails and can be coupled to the operating lever so that every time the carriage slides up, the number is added or subtracted, and every time the carriage slides down, the number is reset into the input. In addition, this carriage can also slide to the left and to the right, to allow for column shifting in multiplication or division.
The actual position of the carriage is thoughtfully made visible through one of two windows at the top, there is a sprung bar in the middle that can be pushed down for shifting left to right, and there si a clearing button that releases the setting sliders for zeroing.
The only thing that the Adsumudi is sorely lacking is a counter register - the user needs to count the number of repeat additions or subtractions in a multiplication or division and write them down digit by digit. This is obviously tiresome, requires concentration, and much reduces the automation of machine calculation. It is easy to see why not many people bought this contraption - it could not do everyhting that the competition could!
Finally, this is a video that gives you an impression of some calculations with the machine. Enjoy!