An inept attempt to hide stolen goods? Or crossing-out a factory mix-up?

To give the old Oliver 3 typewriter a thorough clean, so that it'd be rusty but not dirty, it was taken apart pretty completely. Down to its foundation, the cast-iron base.

The unreadable main serial number on the rear of the base had been attributed to the slapdash black paint-job, but sanding away the black paint revealed that the last digits of the serial look as if deliberately obliterated. It really looks like chisel-blows to the digits.

Removing a serial-number is suspect of course - could it be that this was an attempt from long ago to make a stolen machine untraceable? (And then also given a coat of black paint to make it less recognizable?) 

It seems far-fetched perhaps, but Olivers did get stolen. The Oliver Typewriter Company did keep track of serial-numbers of stolen machines too, as is documented in 'The Battle of Detroit' between Oliver and The Typewriter Trust (Remington e.a.) . A great blog-post about this saga is at oz.Typewriter - fascinating reading!

In case this defacing of the serial-number was an attempt to make the typewriter untraceable, this was a ridiculously ineffective action. Right next to the main serial-number, it's also stamped into the back frame-rod of the carriage. This came out undamaged from underneath the black paint. Did they miss that?

That however isn's the only back-up number that's on the machine. The serial number is also stamped under a tower platform. Not obviously visible, but accessible.

And there's more - normally hidden under the flange of the comb plate; there it is again, stamped into the base casting.

For good measure, also stamped into the main pivot-rod for the shifting.

To top it off, the number is even stamped into the little plate that clamps the leather patch that buffers the space-bar linkage inside the base. After pressing down on it for over a century, the serial-number was also in the leather patch, albeit in mirror-image :-)

Overall, if this destroying of the last digits of the serial-number on the main, raised area for it was an attempt to hide the machine's identity, this attempt failed. Must have been very inept thieves, to even miss the serial number at the back of the carriage. Or there is another reason behind the mauled serial number. 

Could it be a casting-error that made the stamping of the digits ineffective? 

The marks over the number seem however deliberate. Also the chisel-marks would be a strange damage to the (wooden) mould for the casting surely. And in case the raised area were moulded badly, this is a face that could have been easily filed/milled clean again.

Or could it be that the stamping-tools for the large digits and the small digits were not set to the same number?

In case a mistake was made at the factory in 1904, the large number stamped e.g. 114303 and all the multiple small numbers stamped as 114033. Then a reasonable fix would have been to erase the number that was stamped once only; i.e. the large number. Then it'd also make sense to hack away only at the final digits, the thousands' number wouldn't be changed too often and unlikely for an error to be there. In case the mistake was that the large stamp hadn't been incremented, so set to 114032, then obliterating the last digit would have been enough - so perhaps a mixing-up of correct digits in wrong order is the more likely mistake. 

If an error was made in the large number stamp, then the remaining small serial numbers are correct for the machine. However, if the small stamp was mistaken and the serial should have been e.g. 114303, then there would now be a duplicate machine with the same serial-number. (One way to confirm the mix-up theory would be to find another Oliver 114033 - but slim chances of that ;-) Most likely then that the small number stamp was correct and the mistake was made in the large stamp; a typewriter factory in 1904 would be unlikely to create duplicate machines.

On balance; the most likely explanation for the unreadable serial-number is a manufacturing mix-up with a mistake made in the large number-stamp.

Getting a Mignon to write again - feed rollers and adjusting the type

This rusty old Mignon index-typewriter needed some repairs to make it functional again. From browsing online, a fairly common issue with Mignon typewriters is that the feed rollers are 'gone' and glued to the platen. This made that the platen didn't just not feed paper, but it couldn't even be turned - blocked solid.

Removing (and putting back!) the carriage on a Mignon is easy, even simpler than it is on an old Oliver typewriter. Place the left margin-stop between two positions - this will keep the lever of the margin-stop in the up-position.

Then press down on the space-key and/or the carriage-release and simply slide the carriage off to the right. (The space-key and the carriage-release are actually the same function, a 'press' on the carriage-release will space the carriage.)

With the carriage off, unscrewing the plates at the front-sides of the carriage allows the margins-bar and ruler to be removed. Take care to catch the little springs that press the ruler agains the platen. This then gives access to the feed-roller springs that can be removed using pliers - note that these are spring-steel and the forked ends are genuinely sharp!

Temporarily loosening the brackets that hold the feed-roller assembly then allows the complete feed-roller assembly to be wriggled out to the front of the carriage. (I.e. no need to remove the platen; platen is held with pins, pins may not be without rust, etc...)

The feed-roller assembly is held together by double M2 nuts on all axle-ends. When these are removed, it all comes apart. Note that the centre pivot-axle is not symmetrical - there is a left and a right end! Note that also the four connecting brackets are all different and need to go back on their correct position in the assembly.

The feed rollers (now unstuck from the platen) had indeed become 'triangular' and beyond repair. To replace with new rubber, the old rubber was cut away from the core tubes of the feed rollers. Messy.

The 3 front-rollers and 2 rear-rollers reduced to metal tubes, can then be re-covered with a few layers of heat-shrink tubing. The front rollers to approximately 8mm diameter, the rear a bit larger to about 9 or 10 mm. (The Mignon is not very critical on roller diameter :-)

The feed-roller assembly is then put together again - and manoeuvred back into the carriage from the front. The assembly pivot-axle is clamped/held by the side brackets. 

The two curved-springs can be placed back with pliers, to lock behind the notch on the carriage-frame. Ruler and margin-bar replaced and side-plates screwed on to complete the carriage. (And paper-table put back, had taken that off for convenience and to reduce the risk of damage.) Scraping and light sanding of the platen to remove remnants of the feed-roller rubber sticking to the platen, and the carriage again feeds paper!

The carriage then placed back into the machine (easy), it turned out that only the top-halves of characters would print. This must've been an issue for some time, because this machine had the type-cylinder propped-up with a scrap of paper when it was bought. Instead of messing-about with scraps of paper jammed under the cylinder, the type-rod can be adjusted on the machine. This again is fairly easy on a Mignon.

Remove the single screw and remove the cover; press down the two keys to allow the cover to move forward and away with ease. This reveals the core of the mechanism of the Mignon, in all its brilliant simplicity. The large lock-nut holds the large central set-screw that captures the horizontal rod. If needed, adjust this set-screw to make the pointer move freely (yet not too freely). 

The 'flower-wheel' behind the lock-nut adjusts the position of the type-rod and thus the vertical alignment of the type. The little tab behind the rod needs to be pushed back to unlock this wheel, then this 'flower-wheel' can be rotated left/right to move the type up/down relative to the platen. The tab-plate springs back and locks the wheel in position to prevent drift of the vertical alignment during use.

A bit of fiddling and experimenting finds a decent spot for all platen-positions, so that the full character height is again printed. Still a bit top-heavy, but acceptable for now.

With the vertical alignment adjusted and paper feed working too; a new ribbon fitted and this Mignon can write again :)

Comptometer keys - making new

The project Comptometer model H got 'new' keys at its refurbishment in (probably) the 1940-ies, however these cream/green keys of a model J don't age well. Also, the keys of a model H Comptometer should be black/white instead of green/cream. These are one-piece moulded 'plastic' keys, similar to the keys of e.g. an early Oliver typewriter. As part of this model H's current 2023 refurbishment, decided to go for a full set of new keys in black and white. Replacing all keys makes it actually a lot easier, because blending in with any original keys won't be a problem.

In a 3D CAD-program (mesh-editor) new keys were designed, dimensions taken from the least-degraded remaining keys.

The dimpling was of course varied for the even and odd keys, as well as the practical depth of the lettering was worked out with some test-prints.

Also from some (failed) test-prints, worked out a way to make the pockets for the key-stem a bit flexible to get a workable press-fit on the stems. The open areas around the stem-pocket could later be filled after fitting on a stem (e.g. with glue), in case the extra strength should be needed.

Some more experimenting was done to find the best way for finishing the keys. Printing vertical gave the crispest lettering, the visible layering steps in the dimpled key-top could be sanded smooth without problems. Fine-grit sandpaper held over e.g. the ball-end of a file- or screwdriver-handle to press and smoothen properly into the dimple.

Then two layers of enamel hobby-paint to further smoothen the surface and give them an even black (or white) surface finish. How this wears will remain to be seen, but the solid material underneath is at least also black (or white-ish).

For picking-out the lettering, best results were had from filling the numbers with latex paint after lacquering the key with enamel - latex from testers; cream and black latex wall-paint (Black is surprisingly hard to find in latex wall-paint, by the way.)

A whole batch of keys was then printed in black and in white resin. All subsequently sanded, lacquered and lettered to get a full of keys set for refurbishing an 8-column Comptometer model H.

The STL models for a full column of keys for a Comptometer can be downloaded here.

Color-match for repainting Comptometer panels

Coming back to the stripped panels of the Comptometer. The side-panels came out good enough from underneath the brown refurbishment paint, won't be touched. The top-panel was however a (1940-ies?) replacement part without any copper-plating. Also the front display-panel was very worn, so these two top panels were up for repainting.

To select a fair color-match, took a panel to the auto-materials shop. Browsing the shelves with spray-cans, a decent match was found in Motip number 51350. At least, the lid of the can looked closest.

The panels were de-greased, with soapy warm water. The top-front display-panel was sanded down with fine waterproof sandpaper, to remove the rust. It shows all the layers that were worn through, from use. The cork riveted inside the top-panel was taped off.

First a grey primer was applied to both parts. Then two or three fairly thin layers of the metallic-brown lacquer was applied. The lacquer is translucent, so the finish gets darker with every layer. The front display-panel got more layers than the larger panel - and is probably the 'final' color of this brown metallic lacquer.

Metallic paint is fairly vulnerable, so a layer of clear-coat lacquer added to finish the panels. The main panel shows some color variations from thickness differences; translucent lacquer needs a lot of care to get an even finish (and me a lot more practice). Not too bad though, when viewed from a distance it isn't too disturbing.

Overall, the difference with the side-panels is probably smaller than the color differences within original panels. 

All probably too much effort to be put into a pretty common 8-column model H Comptometer, but a fun project - and it'll look nice again :)

An influx of Odhners and parts and tinkering

Because it looked very clean and with the paintwork intact, and it was the full-featured Original Odhner Model 39 version with tens-carry in the counter, back-transfer and check-register, and wanting to experience an Original Odhner descended from what started the pinwheel-type of calculators, took a chance and got it.

It was very well packed and arrived without any mishaps. Dusty on the outside, the internal mechanism was complete and really very clean - almost like new.

Not spotted on the listing images; the calculator had a wrong, stubby replacement crank-handle and a later-pattern left clearing-crank. Minor niggles, but still. Another thing was that the back-transfer did not work reliably - with larger values the back-transfer lever would snap out of its latch and not the full digit was entered into the setting-register. (Transferring a 9 from result would get a 7 in the setting.)

After some exploring, the likely cause was wear to the latch-faces made them slide off under the pull of the (strong) spring. 

The easy fix to try was replacing with a weaker spring. Here the box-of-bits from Remington Portable typewriter EV167240 again delivered! The bell-clapper spring has similar overall dimensions, but thinner gauge wire.

This must be the 6th machine that is fixed or completed with parts from this one Remington Portable :-)

With the numbers from the result-register again transferring correct to the setting-register, the stubby (uncomfortable to use) handle seemed next. 

Just then, a clearly broken Odhner was listed on the local classifieds site - just the thing to use as a parts-donor. Contacting the seller, it turned out that more parts were available - clearing-out of a halted restoration or demo-machine project. Net result; one Odhner purchase leading to another (with correct handles included).

It all arrived in a box in many bags and little boxes - approximately 1.8 calculator :-)

A 3D puzzle - these machines are fascinating puzzles anyways. The amount of interlocks and general ingenuity is impressive. The simple handle of the main crank is even a neat trick to assemble, easy when you know it.

Loosening the screw at the bottom allows the pin to be pushed all the way in. Then the C-clip emerges from the recess in the grip and drops off the pin. The grip can then be slid off the tube on the handle. Then the pin can be retrieved the other way and the spring comes out too. To reduce play of the grip on the handle, regular thin brass M3 washers can be used under the C-clip.

This handle-pin screw was a special pac-man screw - came out fine though :)

Being distracted by so many Odhner parts, there seemed to be enough to re-build a complete model 9, dating from around 1936. Another puzzle and no qualms about experimenting to fix things.

Some shimming of the left frame for clearances and replacing the right side-frame (one was shattered) made it rotate fine and even calculate correct. The button-mechanism for the quick-clearing of the setting-levers did not work and more concerningly, the left-most wheel of the counter register did not clear. This means a broken clearing pin - as it indeed was.

Taking out the register means removing the crank. Cranks are fitted with tapered pins (awful) and require targeted violence with a hammer to remove (scary). This is well described in the section on tapered pins on John Wolff's page on repairs. This pin fortunately came out fine and the register could be taken out, numeral wheels slid off the axle and indeed one broken pin (red arrow). Someone must've really forced the handle when this numeral wheel got caught on the notch-tip, Forcing the handle round, the notch was sheared off and the machine broken - instead of a wiggle with finger-tip of the numeral wheel that would've unblocked it. Oh well.

To repair: the spot drilled out 1.6mm hole with a hand-drill and a nail driven into this hole. The notch only needs to twirl round numeral wheel, so cyanoacrylate glue should be good enough to keep it from working loose.

The nail is then sawed off and the stub filed down to shape. The exact shape and accuracy isn't even that critical; important is that the sharp, chisel-edge tip is at the correct pitch - so 7mm tip-to-tip for this one. After several repeats of re-assembling the register, testing, dis-assembling and filing a bit more - the counter-register works beautifully again on all digits. Fix.

The machine's carriage was also 'blocked'. By the 1930s Odhner was using an alloy for the carriage base that can warp or swell a little over time. This wedges the carriage stuck in its guide, and makes the machine 'broken'. With fine-grade sandpaper, the bearing-sides can however be smoothed and made to fit in the guides of the cast-iron base. After some sanding and new grease, the carriage slides without effort.

The quick-clearing on this specimen probably hadn't worked for ages. This mechanism only works with the cover mounted, the clearing-bar that pushes against the large knob is attached to the cover. A small spring-loaded stud in the drum travels over a curved bracket, to catch it and push the knob-bracket from its latch on the way back. Simple, once you've seen it work. The knob-bracket needed a lot of forming (bending) and some filing to make it clear properly and pop-back.

The counter-direction occasionally does not fully re-set but hangs (red circle); the cause is known (wear) and a solution too (file a slight extra chamfer on cam), but for now this will be left as is. Anyways, giving the crank a little wiggle before a next turn also works to prevent a hang.

As a finishing touch, a comma-bar was mounted back on the cover. Instead of rivets, now fixed with M1.2 screws and tiny hex-nuts (carefully tightened with tweezers).

Not using the quick-clearing function matches the wear of the machine - the paint is worn away exactly where the user's thumb would push back all levers, with the hand resting also on the machine back and on the crank-support of the machine. The serial number 131568 on the back-cover places this calculator somewhere around 1936.

After this tinkering detour, an extra Original Odhner. This one a model 9 pinwheel calculator that looks well used. The worn look actually suits it, so probably this machine won't be repainted or touched-up at all. It again works ok and actually runs remarkably light. 

After this distraction, got back to the model 39 and finally replaced the grip with the correct pattern from the remaining parts. On this calculator the quick-clearing must've worked fine, because there is very little wear to the paint around the setting levers and on the back.

The wear of the lever tips on the right-columns however shows that it was well-used too. It still (again) works fine now. Not quite as light-running as the worn model 9, but very shiny!

Great puzzles to explore, and amazing feats of human ingenuity :)

Clipless paper fastener, all-steel version

The all-steel version of the Ideal Clipless paper fastener.

Simply marked as 'Clipless' paper fastener. No longer marked Ideal or 'CK', not marked that it's made in Japan either. The construction and the same Japanese patent numbers however clearly make this the all-steel successor version of the 'Ideal'.

Noticeable corrosion, green felt can be seen in the seam around the edge of the base.

This remnant shows that the blackened steel base would have been covered with green felt originally.

This seems to be a less-common item. The wooden-cased product is described by American Stationer and there is e.g. a specimen in an Australian collection. There is however very little information about this later version, reported to be introduced in 1922.

This all-steel version is actually not so much steel, even though it does give this impression. There still is a wooden block inside as the base of the machine. The nickelled base and cover are actually made of brass, only the bottom-plate is made of blackened mild steel.

Despite its more industrial, mass-production appearance, the manufacturing-methods used in this device suggest hand-work in relatively small numbers.

After a cleaning and oiling, this Japanese version of the ingenious Bump-method fastener works fine. New, green felt of a similar shade again covers the base, fitted in the same way with a draw-string - elegant, that.

Columbus bilateral letter-scale - repaint and polish

After a repaint and a polish, this little bilateral Columbus letter-scale is very shiny.

This one, like many, looked rather dejected. The brass was severely tarnished from old attempts at cleaning that removed the old lacquer and the cast-iron base was rusty. The pins that hold it all together can however be pulled-out fairly easily to be able to clean-up the individual parts.

The base got to join in the paint-stripping session last week:

The decorations of the base are best visible in the grey primer, just before getting a new coat of glossy black. 

The brass was polished-up, with the remnants of the original lacquer finish removed from the front-face of parts. After polishing, the brass parts were again given a coat of clear varnish to seal the surface as was done originally. Lacquer closes-off the surface to oxygen, stopping or at least slowing down tarnishing.

These small bilateral scales still show up on local classified sites - many were made and they are of course really durable. This Columbus letter-scale was made by the Ph. J. Maul company of Hamburg, Germany. The Maul company made many and diverse scales for over a century - and also made this one, as seen from their trademark on the linkages.

These bilateral scales don't need adjusting against a sloping desk. Because the weights move out symmetrically, it is automatically compensated. This bilateral scale is made to a design that was patented in Germany in 1904. 

This German patent 167,192 was however not by Maul, but by Automat AG in Berlin. The drawings in the patent do illustrate the design, as does the claim of two symmetrical weights rotating on a single axis. 

Possibly that Maul either took a license, or more likely completely took over the production of the Columbus-brand scales of this design from Automat.

Most of these have a brass weighing platform, this one however had a brassed mild-steel platform. The plating was partly gone and the metal rusted. The whole platform was sanded smooth and also painted gloss-black. This is not the original finish but it looks ok again. More importantly, the weight of the platform is not changed too much, so it still weighs correctly.

Perhaps the plated platform is a hint that it was made when brass was relatively costly, perhaps around 1917 to '19 or perhaps during the chaos of 1923. At any rate, it likely is at least a century old.

With small dabs of red paint on the pointers this little device is again good for another century of use. (It has a good chance to outlive the existence of actual letters and postage stamps, but will still weigh things just fine :-)