Layers of history - 'excavating' a crusty Comptometer model F

A not-expensive local pick-up - it didn't look so bad in the pictures (dark, bad lighting). When seeing it at pick-up I did consider leaving it - it was not only rather dirty, but extremely rusty (crusty) and a whopping big iron carrying handle was mounted on the back.

But it was a century-old model F and did look a right 'project', so the sale went through and yet another Comptometer taken on. (Have been quietly acquiring more of these; fascinating machines.)

A solid steel handle riveted to the rear panel. That is not original.

Not subtle, that handle - though it actually is a good idea. It makes it so much easier to pick-up and transport the machine. Even the standing on its front panel isn't too much of a problem, it's fairly stable. This definitely is something that the Felt & Tarrant company could have done themselves (though perhaps not on the rear-panel or quite this crudely).

This Comptometer was handed down from the seller's grandfather, who'd worked with it at Smit, Slikkerveer - that was (is) a large industrial/electrical company founded in the 1880s. Smit was absorbed in Heemaf in '63 and after several corporate moves the Smit activities are now part of Alstom and produce electric-drive for trains. This Comptometer was made around 1918 and likely used for decades at Smit. It probably got its handle added when it was no longer 'new', but still actively used by perhaps a 'roaming' operator. And then taken home at retirement, to be kept as a memento. When it passed to the next generation it ended up in a damp shed. That's when it will have developed its 'patina' - especially the key-stems became very badly rusted. 

When cleaning out the shed of the father, the grandfather's old machine was put up for sale - there might be someone that'd still want it.  There was :)

After getting it home, it was first stored in the garage (much too dirty for taking indoors), before taking it out for a proper cleaning and repair over a a few days - an 'excavation' almost. 

There are minor differences between the models, but all 'shoebox' Comptometers can be taken out of the casing pretty much the same way. This gives access to the mechanism for a cleaning - this one was dusty even on the inside. The empty casing can then safely be given a good wash.

Fortunately the mechanism itself didn't need taking apart much (intimidating, quite beyond me). With liberal amounts of a light machine oil (sewing machine oil) and working the various parts, the whole mechanism slowly came to life again. At first a bit slow with 'sticky' carries, but after a while the full ripple carry from 099999999 to 100000000 again zips through the machine. Everything works and no broken or bent internal bits!

With the top-plate off, this also could be thoroughly scrubbed. One thing I'd not noticed before on Comptometers was that the key-stems had a gasket; every stem-slot had a fibre or card sleeve to act as an oil-soaked gasket. These were hard with old oil and/or crumbling and were all removed.

Considered replacing these with new, but did not want to risk one of these gaskets coming unstuck and then getting lost inside the mechanism. With the light use this machine will see, it should be fine without these oiling-gaskets. (Would the fact these gaskets were still present suggest this specimen hadn't ever had its top-plate removed for a full-overhaul maintenance?)

Another layer of its history were marks from a divider placed between column 2 and 3 (cents and guilders) and between row 5 and 6. The copper looks scratched/worn away by a divider, down to the plain steel. 

The horizontal mark suggests a divider to help practice/use the touch-method. Experienced (and fast) comptometrists only used the 1 to 5 keys in every column. All numbers were entered by touch without having to move the hand, never taking the eyes off the figures to be added made this a very fast method.

It was also revealed that the greasy dirt-layer between the keys had actually been protecting the copper finish of the machine - still closer to the original colour than the rest. The darkening of the 'clean' surfaces is probably due to decay of the clear lacquer and oxidation of the copper underneath.

Even though it was part of the machine's history, the rusty handle on the back was removed. The holes left by the 1/4" rivets (!) were plugged and made less obvious with copper-colour paint. The lighter, original colour where the handle had been also confirmed that the darkening (and rust) was probably from its time in the shed.

The patent dates again all readable, from the original 1887 to the most recent 1914 date for a model F.

All the stems were scrubbed clean with steelwool and the cellulose-nitrate keys were washed with soap and water. Kept in sets per column, to place them all back in their original spot. Bent stems (several!) all carefully straightened.

The case was cleaned with soapy water and 'crusty' areas were polished down to a smooth rust-finish. The case now has multiple shades of brown in interesting patterns that would be the envy of some modern 'industrial-look' decorative items. Sealed with a waxing, safe to touch and handle.

With clean keys and remarkably shiny key-stems, it all looks much better :)

The re-mounted, clean keys confirm that this machine was used by an experienced operator - the lower half of the keyboard is worn. This shows it was used mainly for addition of amounts up to 4 or 5 digits large using only the lower half of the keyboard with the right hand - fingers resting on the 3-keys, as per recommended method.

The machine was completed by re-fitting a red correction-key; again a proper well-used Comptometer model F. 

The model F was introduced in 1915 with a white Correction Button for re-setting the controlled-key mechanism and blank decimal-markers. Numbered-markers were available sometime before mid 1917 and the correction button had already changed to red before that. With serial number 135855 this Comptometer was probably made in 1918 so will have had numbered decimal-markers and a red key from new.

It was actually surprising how well this machine could be recovered back from its extremely rusty 'boat-anchor' condition. A nice (and cheap) addition to my modest collection of Comptometers :-)

With occasional 'exercise' this Comptometer should remain in good working order for decades to come as a 'collector-item' or historical technology-artefact; a new layer being added to its accumulated history.

Why do people keep wrecking these boxes?

Why is it, when a drawing-set case is in a thrift-store, the latch of the box is always destroyed?

Probably because nobody remembers these cases. I.e. that these have a little latch at the front-right corner (or sometimes have two, at both front corners). Pulling-out the latch will unlock the box and -voila- the lid can be opened. For anyone encountering one of these baffling boxes; that's how they open.

This was a decent late 1930s to perhaps 1950 set, made by the firm of E.O. Richter in Chemnitz, Germany. A medium-sized set with ink-pen, a large compass, extender and a drop-bow compass. These are pretty common, but are quality instruments; worth trying to fix. 

To repair the latch, an old 2mm knitting-needle was filed down replace the missing section. The combined part (cyanoacrylate) makes the assembled 'rod' have again the slot for the little nail that locks it in orientation and limits its travel in its channel in the wooden base.

After re-fitting the latch (slide in from the right), placing the locking-nail (through the slot and push firmly into the wood), the purple box-lining is carefully glued down again as is thin paper-edging. An almost invisible repair, this.

These larger Richter-type compasses have a little mechanism that keep the handle straight - this must not be forced; you can't open the legs by pulling one leg and holding the handle. If you do, then the small guiding-rod snaps. That's probably another bit of once-common knowledge that's no longer all that common: the compass should be opened by holding the legs, not the handle. Thrift-store: so a broken guiding-rod. This little guiding-rod is now replaced by a new 'rod' fashioned from a length of M1 threaded rod and two (relatively massive) M1 nuts. (In picture below, centre, they're really rather small :-)

Yet another thing with thrift-store drawing-sets is that bits go missing. The case gets opened (broken open), handled and something will accidentally drop out and is lost amongst the rest of the jumble. This set is only missing its needle-attachment for the large compass. I have a spare part for this, but no needle-screws yet, so this is something to hunt for (or re-create).

These fairly-common Richter sets are difficult to date reliably, but the black tapered plastic handles would make it not earlier than the mid 1930s. The old style of Richter-brand and absence of any GDR quality marks suggest it is earlier than 1950 and probably before 1945.

In any case, the latch-repair turned out well and even though these sets are common (and not expensive), it is a very nice and still usable drawing-set. (And at a fiver, no regrets about getting it :)

Standard Folding 2 Typewriter - restored

The barn-find from Jubbega; restored:

Front panel lettering of the Standard Folding Typewriter

The very distinctive 'Standard Folding Typewriter' lettering on the aluminum front-panel is now also applied to the restored machine. At first it was a bit faded and not quite the result that was aimed for, but it did look 'aged' and in keeping with a century-old machine :)

In the process of restoring this derelict machine, the front panel was completely stripped to bare aluminum again. To complete the restoration, it was also to get a new 'decal' with the distinctive 'Old English black' typeface stating that it is a Standard Folding Typewriter.

To start with, a suitable 'Old English' typeface was found and the basic lettering created with it. This already gets close, but there are still noticeable details that are 'off'.

Using some of the many photographs that are available online of Standard Folding typewriters, the lettering was tweaked to better match the letter-kerning and line-spacing.

Then the size of the lettering was estimated from counting pixels on photographs, and calculating what percentage of width of the front panel was the distance between the outside of the uprights of the 'F' and the 'r'.

This lettering was then laser-printed mirror-image on silicone-covered backing paper. That's by re-using the backing of a regular label-sheet for laser-printing - giving several 'heat-transfer' labels.

A 'transfer' was then taped to the front-panel and an iron was used to heat-press the black laser-toner onto the aluminum surface.  The sintered toner again becomes soft and adheres better to the aluminum than it does to the backing-paper, at least that's the theory. (In hindsight, should have cleaned the aluminum surface better, with acetone or similar de-greaser - a next time...)

Some corners of the decal somehow didn't transfer neatly - the right-most 'r' was e.g. missing. Also the black toner transfers to the textured aluminum imperfectly, as if leaving the tops of the texture exposed and resulting in a faded, grey appearance. (Possibly some printer-setting could be found to apply excess toner, again something for a next time...)

The net-result after application was not quite what was hoped for; it does match the look of an aged machine, but does not match the 'new' keys and overall clean machine. On the plus-side, the heat-transfer laser-toner is not fragile - it's a very rub-resistant 'decal'.

Alternatives could have been a waterslide transfer or direct ink-jet printing. Because of the bare, textured aluminum a waterslide transfer is however not ideal, because the outline of the backing would probably show clearly. The direct inkjet-printing onto the aluminum surface also has its challenges; a consumer-printer would require some significant modification to accept the front-panel. A more suitable printer would be a UV-cured inkjet printer that can print also on irregular surfaces. These are usually found in a professional setting (e.g. as used to make promotional merchandise, printing on bottles etc.), so also not readily accessible for tinkering/exploring as an option.

With some careful retouching, the end result of this attempt with heat-transfer of laser-toner is however not too bad. There were enough 'hints' from the transfer to create the missing 'r' and to e.g. complete the 'd'. (Retouching was done with a fine camel-hair brush and black latex paint - this paint can be easily removed with a damp cloth in case of mistakes. Not so durable, but safe and reversible.) 

End-result is a fairly decent reproduction 'decal' on the Standard Folding Typewriter, fitting for a restored 100+ year old machine. 

(There is always the option to clear the panel with a gentle glass-bead blasting and re-do things; in case a better option comes along. For now, we'll leave it for a bit - see how this 'ages' :-)

Imported from St. Petersburg to Rotterdam, an Original-Odhner arithmometer model Ag

Imported from St. Petersburg, Russia to Rotterdam, Holland around 1908 by the J.A. Ruys trading company,

The Ruys trading company was the continuation of the Voorhoeve trading company, renamed in 1904 when J.A. Ruys bought-in to take over the company. A major line of business at that time was the importing and sale of Hammond typewriters, and Odhner calculators as shown in the 1904 newspaper ad below.

Hammond machines in The Netherlands often have the brass Ruys shield in the same style as the Hammond shield on the wooden base.

Ruys also advertised the Odhner calculator by itself, for example in this newspaper ad from October 1907.

(More advertisements and some background on Odhner at the informative mechanical calculators pages of Jaap Scherphuis.)

The St. Petersburg-made Odhner machines really are the original pinwheel calculator, the start of the 'category'. As such, it is a historical milestone-machine for automated calculation When this one showed up on the local classifieds site, curiosity won-out and I made an 'impulse purchase' - pretty much the item as advertised in that 1907 newspaper ad.

With almost 30,000 made, these calculators should not be rare and be available at relatively modest prices. The very early ones (with short crank) are however eye-wateringly expensive these days and even the later, nice specimens with case are becoming sought-after. This later specimen in the state it was in, was fortunately affordable.

This is a model A with 13 digits capacity in the result register. An extra 'g' indicates a bell, so this is a model Ag. The marking with 'Original-Odhner' instead of 'Odhner-Arithmometer' started in 1907 (to set it apart from the increasingly successful clones). Going by the serial-number of 14618 and the advanced features (comma-sliders, fast-clearing bar) this machine was probably made in 1908 or perhaps late '07.

It had been given as a present (retirement gift?) to the grandparents of the seller. Nothing however was remembered about the background or the occasion, it had just been 'around' for decades. There was no case and it is mounted on a (1950s?) small multiplex board. And now sold.

The calculator was also missing its bell, several screws and generally rusty and dusty. It was also very stiff - so did not try to operate it before cleaning and lubricating. Otherwise there is risk of blocking the machine or even breaking something from excess stress.

Without taking apart any of the more complex sub-assemblies, like the drum or registers, it was laid out in its main components. All mechanisms were oiled (sewing-machine oil) and gently worked free. Some assemblies may have to be taken apart later for more thorough cleaning and removing excess (and old) oil.

Metric M2.5 was a good fit for replacements for the cover-screws. Heads were modified to match the pattern of the remaining original screws ('instrument-head'). 

Unsure what screw-threads were standard in pre-revolution Russia, but when the Odhner production-design was made in 1890 or so it was probably designed around 'Sellers' or American threads. The cover-screws are what today'd be called UNF #3-56 screws - the M2.5 is nearly identical in diameter, pitch and has the same 60 degree thread-angle. (International metric is derived from French standard screws, that are in turn derived from William Sellers' American screws with 60 degreed angle.) 

Assuming old American-size threads, the screws in the register wing-nut flange are probably #5. Anyhow, a decent replacement for a missing screw was scavenged from a 1947 Underwood standard.

Most notable of course was the missing bell - a small bicycle bell! At 32 mm diameter it is slightly too large, but does fit and at overflow makes a wonderfully bright 'Dingg!'. It being black also is in keeping with the machine's overall 'not shiny' appearance :)

On older Odhner machines without bell, the carriage can simply be slid out. On this model however, a locking screw has to be removed first (and the bell taken off too). This screw is accessible by first removing the back-panel - it has a squared-off head that will also prevent it working loose as it strikes the two end-stops.

On the right in the image above the only safety interlock of the machine can also be seen. The steel disk with one cut-out works in tandem with the slotted bar to prevent carriage and drum being operated at the same time.

The machine is now free from the copious dust and dirt. The mechanism is mostly functional, although not yet as it should be in all positions. Some of the register-wheels hardly want to move - more disassembly and cleaning will be needed for that. There is also the chance that it is just worn-out. 

Additionally, some of the gear-train and handle have developed play and cause the timing to go slightly off for the safety interlock - to be looked at later too. The rusty covers could perhaps be re-finished, but it may all be left as it is. The calculator just looks its age :)

It also definitely looks 'old' when compared to a 1930s Odhner, side by side in the above image. The later, Swedish-built machine is smaller than the older Russian-made machine, yet with many more features and safety interlocks. The size of the old machine, its features and also very much the curly script do evoke the turn of the century (the 20th century, that is).

A survivor from the dawn of widespread mechanical calculation :-)