Awkward adding machine repairs, a Dalton that's not a Dalton

As an entertaining puzzle. Originally this blocked-wreck was bought to be a source of American-size screws and some spare parts, but was also tempting to see what can be learned from it and what could perhaps be fixable.

This is a Dalton, but it is also not a Dalton adding machine. It was indeed manufactured in the old Dalton factory in Norwood, but this is not a Dalton design. This really is a Remington-Rand 'Portable' adding-listing machine. The Remington adding machine is actually the Brennan adding machine. This company had a brilliant, compact design by Thomas Mehan, but had the misfortune to launch new (expensive) business machine in 1929 just before the crash. When the Brennan company went under in the early 30-ies, it was bought by Remington-Rand. 

Remington-Rand had itself been formed a few years earlier in '27 from the merger of Remington, Rand-Kardex, Dalton and Baker-Vawter Ledger. The newly acquired Brennan design was superior to the by-then aged original Dalton design and from '33 the Brennan-design was basis for the new Remington-Rand  'Portable' adding-listing machine.

Most of these new 'Portable' adding-listing machines were sold as Monarch, many as Remington and some as Torpedo (Germany) or Dalton. It seems the Dalton brand was used mostly in export machines, e.g. in regions where the Dalton name was still valuable. This particular 9-column specimen with display, serial number M 150,893 is from 1937. Like many late 30ies Dalton's, this one was also originally sold in Belgium.

A lot of tweaking and oil on several cams un-blocked the main movement. The blocking was caused by hardened old oil/grease stopping lever springing back when they were expected to do so. The keyboard was however badly rusted, the key-stems no longer moving freely in the slots in the equally rusty key-plate. This specimen being a wreck anyways, the keyboard was taken off and taken apart.

These 'Portable' machines are not easy to work on, the keyboard-assembly is really hard to get out (as commented on by bss1250 in his video on a Monarch). Some bending of the outer side-frames is needed to get it out. The bottom-plate can be screwed off and then the routing of the keys to the number-positions is visible - the row of 'prods' that enter a number into memory are at arrow A. The protrusions on the line B are what trips the universal bar (itself in top-right of picture). All the stems C have a delicate spring on it that pushes the key back up.

When the keyboard is removed, the 9-column register can be seen on the machine, slidable on its rails. The teethed/castellated rack in the front of the memory-bank (pin-bed) engages with the escapement, to index it one position on every digit entered. The pin that's pushed down in a column enters a number, just like the original Dalton. (That lever in front was a bit worrying, that fell out when holding it upside down! Took a bit of puzzling to find where it should go - it's the lever that prevents keyboard-clearing when the repeat-button is pressed down.)

All the key-stems are uniquely shaped, they were all cleaned with steel-wool. Having gone this far, the key-plate was stripped and re-finished with dark green paint.

Re-assembly of the keyboard-assembly is 'hard'. The 13 delicate springs need to be held in place on 13 keylevers that themselves need three tabs to be in slots of both top and bottom plates. The hole in the 'prod' of the key-levers now makes sense - this allows a wire to be fed through to keep them in-place during the assembly. Only after several attempts and extra spring-clips to hold things together, was the keyboard put together again - this is not meant for maintenance!

Again bending the side-frame and a lot of puzzling on how the various levers have to be positioned, the keyboard could be screwed back onto the machine. Now with clean key-stems that spring-back easily.

Numbers could then be entered into the machine, the adding mechanism however still made mistakes. When adding 15 to 15, the result was 20 - none of the ten-carry's worked. Adding 3 to 7 made zero.

How or where the mechanism does (or ought to do) the tens carry was not obvious from peering at the mechanism. Reading Mehan's original patent for the mechanism and looking at the drawings gave however a great explanation on how the carry's are supposed to work. (This by the way also made understandable why an 'empty-stroke' is always needed before the total can be printed.)

When a number-wheel (71) passes from 9 to zero, a notch on this wheel (103, marked green) pushes the lever 104 down, allowing the pin 56a on the rack to enter the slot in lever 104 (circled red). The lever 105 on axle 108 (marked red) needs to spring forward to hold 105 down for the carry to happen cleanly.

On this machine, all levers 105 were glued solid on their axle 108 (red arrow, below). This prevented any carry from happening.

With fresh oil and by forcing the levers to move to break the hardened grease, the levers 105 again moved freely. This actually fixed the carry's!; adding 15 to 15 again gives 30.

With the keys also cleaned (they were grimy-black all-over), the whole machine might actually be fixable.

Viewing from the front, the indexing mechanism is blocking the view to the pin-bed memory behind it. The universal bar (white arrow) is what trips the escapement, the long horizontal spring at the bottom pulls the 'carriage' of pin-bed memory to the left. One position at every digit entered.

These Remington small adding machines are not rare and not valuable today, so a good candidate for the occasional puzzling to fix all the other things that still don't work (e.g. the printing mechanism insists everything is a sub-total). And in case of failure and breaking it, not a big issue either.

A 3D crossword-puzzle, as it were ;-)