Marchant Calculating Engine

This Marchant "Silent-Speed 8M" was aquired by the observatory sometime around 1936, for $600. It was made by the Marchant Calculating Machine Company (which merged with Smith Corona in 1958 to form SCM Corporation) of Oakland California. It was used (along with a Marchant ACT10M, purchased around 1942, for $750) until the late 1960s, when these calculators where replaced by the Olivetti Programma 101 programmable calcultor.

The numeric input keys are arranged in columns, to allow the operator to input the numbers to be operated on. The accumulator/display is mounted on a typewrite-like carriage that moved back and forth during multiplies and divides. Divides involved successive subtractions that could take a long time, so the DIVIDE STOP key was important in case an error was discovered before the calculation had completed.

This device accomplished the tasks of addition, subtraction, multiplication and division by rotating a series of cams and wheels clockwise (for addition or multiplication) or counter-clockwise (for subtraction or division) the required number of times. This was initially done by turning a crank manually, and the major advantage of the electro-mechanical Marchants was that the cranking was automated, saving the human calculator from a great deal of repetitive cranking. The basic problem of the design of a mechanical calculator was how to move a gear an amount proportional to the number to be added. The simple stylus/slide adders had an easy answer to this. The user simply placed the stylus in an appropriate hole and the wheel or slide was moved by the appropriate amount. This was undesirable, however, because there was no way for the user to be sure that the correct number had been entered. Also, in the case of multiplications and divisions, the same number would have to be "dialed in" over and over. These problems lead designers to produce more complicated machines.

The mechanism used two disks which were pushed apart by a spring. One disk had 5 equal width cogs along part of its circumference. The other disk had 4 cogs of varying lengths.

In the picture, the two disks are in the zero position - neither set of cogs will engage the counter wheel (blue). The disks were squeezed into position by two arms which were squeezed the appropriate amounts by the keys above them. A top view of one set of arms is shown below with a side view of one of the keys to the left.

Keys 1-4 caused the right (lower in the picture) disk to be squeezed towards the counter gear to place 1 to 4 cogs in its path. The 5 Key squeezed the left disk towards the center which placed its 5 cogs into the path of the counter wheel. Keys 6-9 squeezed the left disk in as did the 5 key but also squeezed the right disk in by an appropriate amount to place 1-4 additional cogs in the path of the counter wheel.

Counter wheel

With the problem of entering variable numbers solved by one of the above methods, the counter gearing was a rather simple affair. The only real challenge was to make them step into place "digitally" rather than moving continuously like clock hands, and keeping them from stepping too far due to their own momentum.

The above picture shows a typical counter in an Odhner style machine. The pinwheel drove the green gear which drove the blue gear which was attached to the black disk with numbers printed along its rim. (On stepped drum machines, the numbers were usually printed on the side of the black disk opposite the gear.) The gray click stop prevented the counter wheel from overshooting during fast cranking which was a major problem with earlier machines. Even with the click stop shown, the gears could sometimes be overthrown when the machines were operated in a jerky fashion, so more elaborate stops were devised. They made it impossible for the counter gear to move without a full back-and-forth pivot of the stop. (Somewhat like a clock escapement.)

Carry Mechanism

In most calculating machines, all of the digits of both numbers were added at the same time. Regardless of which of the above mechanisms were used, turning the crank caused all them to go into action at once. This was fine except for the problem of carry. Carries had to be handled from right to left after the addition, and the act of adding the carries could necessitate additional carries. The reason the cogs on the pinwheels or drums only went a portion of the way around the circumference was due to the room needed for the carry mechanisms.

The following picture shows part of the carry mechanism on a typical Odhner-style machine.

(This is the other side of the counter wheel shown above.) On a typical Odhner machine, when a counter digit went from 9 to 0, a wheel (orange) with a single tooth pivoted a hatchet-like projection (gray) out in the path of the next pinwheel to the left. The end of the hatchet had slanted surfaces (light gray) and was thin at the top and bottom but thick in the middle. (See side view inset.) This device had a click stop of its own and stayed in place until it was pushed back.

In addition to the nine digit pins, each pinwheel had two more pins (one used during subtraction and one during addition.) Rather than moving in and out, these pins were always out but they moved from side to side. In their normal position, they were out of line with the number pins and missed the counter gear. However, the projection from the counter assembly in the next position to the right pushed them in line with the gear so an additional unit was added.

There were two sets of pins because on an Odhner machine, the pinwheels were turned in one direction for addition and the opposite way for subtraction and the carries (or borrows) had to be performed after the addition or subtraction. Also, the carry pins were staggered around the pinwheels so that the rightmost digit was handled first and any carry caused by the previous carry could be added to the left. After the caries were performed, another part of each pinwheel pushed the carry devices back into their normal places.

Other Complications

In addition to the above, mechanisms were needed to zero all the registers. Sometimes a mechanism was included to zero the levers as well. (On ten key adding machines where the number was cleared on each add, a key and mechanism was provided to prevent this while doing multiplications.) Bells were typically provided to indicate underflow and overflow and a great deal of interlocks were typically provided to keep the user from making mistakes including:

• A mechanism which prevented the levers from moving when the crank wasn't in the rest position.
• A mechanism which prevented the crank from rotating when the carriage wasn't properly aligned.
• On non-Odhner machines, a mechanism was provided to ensure that the crank was only turned the right way.
• On Odhner machines where the crank was turned in one direction for addition and the other for subtraction, a mechanism was provided which forced the crank to be turned completely in one direction before allowing it to go in the other. (The carry mechanism would do the wrong thing if a complete turn was not carried through.)
• On keyboard machines, mechanisms to prevent multiple key presses in the same column or partial presses of keys.

The earliest attempt to build an electronic computer was by J. V. Atanasoff, a professor of physics and mathematics at Iowa State, in 1937. Atanasoff estimated that it would take eight hours to solve a set of equations with eight unknowns using a Marchant calculator, and 381 hours to solve 29 equations for 29 unknowns. The Atanasoff-Berry computer was able to complete the task in under an hour. The first problem run on the ENIAC, a numerical simulation used in the design of the hydrogen bomb, required 20 seconds, as opposed to forty hours using mechanical calculators.

References
[1] From the excellent Museum of HP calculators (section on How Calculating Machines Work).

Return to Hall of Precision Astrometry

Last modified Wednesday, December 28, 2005 © 2009 by The Rectors and Visitors of The University of Virginia Maintained by the webmaster