Thales CES
The very special Thales model with the extra memory at the bottom of a regular CER machine. Useful, I'm sure, but also expensive at the time, and thus quite rare. The big problem with this system is that the memory also has to carry the tens correctly, and thus, apart from the coupling mechanism between the result and memory register, the entire base is filled with a tens' carry mechanism. It operates quite ingeneously - while pulling the large lever forward clears the result register and transfers the numbers to the memory, it also winds a spring, that subsequently rotates the tens' carry cylinder and effects the tens' carries as the lever is returned to its rearward position.
I wanted to put this machine online quickly, despite it still needing restoration - the machine has been badly repainted in the 1950s, and should get its original black crinkle finish and regular black paint for the covers back. In addition, all the comma sliders and comma slider bars are missing. Mechanically, the machine was blocked, but this was easily rectified. These pictures can serve as "before" pictures for the restoration:
Mechanically, in any case, it is fine, it is just the colour which needs a little help.
This machine is serial number 31091 - the only other machine known to me (on rechenmaschinen-illustrated.com) is nr. 31105. One wonders - did they start with 31000, and how many were made ?
In the mean time, I started the restoration of the machine. The paint on the covers was stripped, and the the brightwork polished. Inside, some cleaning was done as well. This is the state of the machine as it is now:
The paint has been stripped off the covers and most other overpainted parts, and they are ready for primering and repainting in gloss black. A very nice surprise was that the back plate of the machine had not been ground down to blank metal as the rest of the covers had, but retained its original coat of black, overpainted with hideous battleship grey. Since Thales' original black paint is clearly tough as nails, the paint stripper stripped the battleship grey off, and left the original paint completely intact. With a light polishing it is fine, so the machine retains at least some of its original paint. The inside of the cover has a cross scratched on it to show what the inside is, and also scratched are the two last digits of the serial number - 91.
I have undone the machine from the base so that I can put paper in between to cover everything that does not need to be painted in the colour for the base, which is a wrinkle finish black. Wrinkle finish paint can still be had, but it is not easy. I've gone with Hycote spray paint, which I've never used before, and which has to come from the UK because I cannot find it locally. The "local" brand, VHT, is not being sold at the moment pending EU approval of the new formula.
While waiting for the wrinkle finish paint to arrive, I've painted the covers and filled the text in with off-white, a Levis latex paint tester which I've had for a decade or so. Latex paint is good to use for this purpose because it dries fast, and while semi-dry can relatively easily be rubbed off, leaving only a residue in the cutout letters and numbers.
The wrinkle finish paint arrived, and called for two light coats 10 mins apart, 4 hours drying time, and then a last, thick coat, which would wrinkle up. I must say it absolutely did work as advertised. However, due to putting the paint on too thick, I have a few unsightly runs on the vertical surfaces. This could probably be remedied by keeping each of these surfaces horizontal in turn, and paint the machine in five installments. I'm not sure, even now I know what it looks like when you do it in one go, whether I would give that a try the next time. In any case, the wrinkle effect seems to have worked properly.
After this, I needed to scrounge some screws from a parts machine I have, because plenty of the holes in this calculator have had "self-tapping screws" installed of larger diameter than original (seems to be 2.8 instead of 2.5mm). Where possible, I took these out and replaced with original size.
And finally there were some trials and tribulations with the functioning of the summation register - it only has two controls - the large lever on the left hand side, which clears the result register in the carriage and adds or subtracts the numbers to the summation register. Addition or subtraction is controlled by the pair of small levers on the right. Down is for addition, and there is only one possible setting for the lefthand lever when in this position. I suspect that if everything works as it should, it should be possible to place the lever in "subtraction" mode with the machine at rest, perhaps with the help of the tiny nickeled lever at the back right of the base. However, something jams. The levers on the summation register can only be moved when the large lever for transfer is pulled forward very slightly. Once both levers at the summation register are up, there are two possibilities for the small left hand lever - either it is blocked in the downward position, and then the right hand lever stays where it is through a number of transfers, or, when done with a single correction, the small left hand lever is pushed to the upward position, and on the next stroke of the summation lever, the register will switch to addition before the stroke commences. Obviously, it is impossible to switch the right hand lever back to addition with the machine at rest, or there would be no need for this elaborate procedure.
The item counter will always count up, regardless of the setting of the levers on the right, and as far as I can see from the mechanism, there is no provision for it to count in the other direction.
The mechanism that controls the addition or subtraction is rather ingenious. There are two axles with relatively broad gears. The Add/Sub. levers control which of the two axles and gears shifts to the left to engage with the intermediate gears at the front of the carriage, which are visible through the carriage top plate.
If the bottom axle shifts to the left, it shifts enough to engage with these carriage gears, but th gears on it are broad enough to also stay in engagement with the next bank of intermediate gears deep inside the summation register, which in turn is always in engagement with the numeral wheels of the summation register. So the item count goes from the numeral wheel in the carriage through the carriage front gear through the bottom bank of gears through yet another bank of intermediate gears to the numeral wheels in the summation register. Three intermediate banks of gears mean that this is the subtraction mode - both numeral wheels turn in the same direction - if a 2 transitions to a 0 in the carriage, a 0 transitions to an 8 in the summation register. This is all completely impossible to photograph when the machine is assembled. The famous bottom axle can be seen in the following picture, right below the top axle which we have not discussed so far:
In addition mode, it is the top axle that shifts to the left and engages with the front carriage gears. In these pictures the axle has shifted left:
And back to the right after the end of the operation (don't mind the item counter going from 4 to 2 - I picked the 2 best pictures to show this from more or less the same angle, they are not from the same operation - it should go from 4 in the first picture to 5 in the second!):
Since these gears are also wide, they always stay in engagement, regardless of their position, with the gears on the bottom axle, which themselves are now out of engagement with the carriage front gears, but are still engaged with the next intermediate shaft. The net effect is a reversal of the direction of the summation register, so that a 1 transitioning to a 0 in the carriage now transitions a 0 to a 1 in the summation register. What about the top bank of gears when it is not shifted left, I can hear you think (in case I haven't lost you by now)? It indeed always remains engaged with the bottom bank of gears, and turns along with it regardless of the position of either of them. But in subtration mode, it is not itself engaged to anything else, so it is just turning along for the ride with no effect whatsoever.
At some point I will draw a colour-coded diagram of the gear train that makes it clearer. For now the pictures will have to do.
There were some worries about tens carry in the next to last digit of the summation register, but exercising the tens carry pins in the carry cylinder, which were a bit stiff, solved this issue.
The funny bit at the bottom plate to the right of the carriage (see the second picture below) is a bank of gear wheels with a number of teeth removed in its rest position. They rotate along if the large lever on the left for result transfer is used, and pull along the clearing in the carriage via an intermediate gear. The missing teeth allow the carriage to be cleared when the large transfer lever is not being operated. That there is a gear in every position hints to the fact that result transfer to the summation register is possible in every position of the carriage, not just the rest position - and that is indeed the case. What it would be useful for is another matter entirely.
That is as far as I have been able to figure things out. If anyone is in the possession of a manual for one of these machines, I'd be happy to hear about it!
One final touch were the comma sliders - they were all missing, but I had a parts machine, which was a later CES. Of course, while the sliders themselves are identical, the length of the bars is not the same over a 20+ year interval and a thorough redesign. I looked for steel bar of 1.5 x 3 mm, but this was impossible to find, so I opted for brass. The slider bars are now suspiciously gold-coloured, but this is not that obvious when they are very brightly polished (although now of course I told you ...). At some point, if I have a batch of things to be nickel-plated, they will go along to the plater's and the machine will really look as good as new.
Finally, some pictures of this very rare, almost fully restored machine:
Enjoy!