Experiencing an album - centuries

An album with classical music taken out this weekend.

This one a Columbia issue of Schubert Trio, Opus 99, dating from 1927. The 4 records are in a sturdy album, tied with a ribbon. 

The gold printing on the cover has some flaking/damage, but the album is fine and the records are in excellent shape.

Listening to the entire work means playing the eight sides in their order.

Playing a side starts with winding up the gramophone (~30 turns), placing a new needle, carefully placing the record on the turntable, wiping the surface with the brush to remove any stray debris, start the motor and then gently place the needle on the record.

Then listen to the side, reading the sleeve commentary.

When the side is finished, lift the reproducer off the record, remove and discard the needle and turn over the record - or return to the sleeve and take out the next disk.

Listening to a classical album this way is not a 'hands off', button-press activity - between every ~4 minutes of listening there is the changing-sides ritual.

The introductory note on the first page gives background; that it was composed by Schubert in October 1827. This music was a century old when this performance was recorded in December 1927 in London.

Placing the needle on every side in 2026, the records themselves are now almost a century old. That realisation did add to the listening experience, amazing and wonderful that this is preserved and still playable.

This 1927 recording is today in 2026 availabe for listening online too! Also amazing and wonderful. 

In the online 'capture' of the recording there is a small gap or silence every ~4 minutes where the performance was split over the multiple record sides. For example the switch to side two of part one is at 3:48 in the full piece available on Youtube (or The Archive).

An enaging and enjoyable music listening experience - spanning a few centuries :-)

Comptometer E - part number J281 celluloid window

Perhaps the celluloid window-sheet of a Comptometer was part number J281. During cleaning and repair of my Model E Comptometer, noted a crisply embossed number:

The window sheet may well be original, although this Comptometer clearly was serviced and repaired in the past. So it may be a replacement, but probably done by a (the) company servicing - perhaps in the 1930s using a Model J part? Maybe the spare-parts lists, drawings or Bill of Materials of a Comptometer will turn up in the future. 

The snippets from a repair manual of 1919 on Brooke Boering's site confirm that Felt & Tarrant used three-digit numbers, some with a letter pre-fix. For example on a Model H the subtraction (cut off) lever is part T604, a toggle pivot link shaft is part 865. Perhaps we can now add that the curved celluloid sheet is part J281.

The Model E was a much more complicated machine than its predecessor C. With this model Felt and Tarrant added error-detection and control mechanisms that not only increased cost, but also made it relatively fragile. (These new features would likely have been considered essential by the company in their fighting off the direct competition of the Burroughs Calculator that launched 1912.)

The Model E machine's frame is noticeable thinner than the preceding Model C, comparitively flimsy. The keystems with the many corners (stress concentration at bending) and narrow sections are vulnerable too. This specimen indeed had a broken key-stem and the top plate had lifted from the holding brackets at the front. This distorts the case and can jam the cut off levers.

Even though from the classifieds ad it was clear it was blocked, had a broken keystem and was rough (and dirty!), it also was 25 Euro and a Model E: so did a pick-up of the machine. Model E's are not all that common, that likely is because the E is not as robust as other Comptometers - it is rather fragile, in fact. Even though thousands were made (7700?), the survival rate of the E is expected to be much lower than that of e.g. a C or an F model.

To clean and repair, the usual steps for a Comptometer; the mechanism was taken out, all keystems removed and everything was cleaned. Taking out all keys enables cleaning also of the top plate in the key-area, often has the accumulated dirt of many decades. The Model E key-stems can only really be cleaned when out of the machine. 

The broken 9-stem was given a 'fix'. It really should be hard-soldered or brazed, but not (yet) having the tools for that it was first soldered. To compensate a bit for the soft joint and anycase to mitigate the butt-joint a splint was added. Slots were sawed in both ends and a sliver of iron inserted. (The lower end of the broken keystem was initially thought to be missing - simply could not be found in the case or in mechanism around column 9 - only when re-inserting column 6 did this bit drop out with a tinkle. Lucky!)

The soldered joint then sanded smooth and the key can be fitted back into the machine. This weak keystem not placed in its original corner position of column 8, but swapped with the intact stem of column 7.

This joint is too soft really, but should hold for a few years until a proper hard-soldered (or welded) repair can be considered. 

Again complete and fully functioning - when taking care to operate gently. Re-assembly also needed care, the order of tightening of screws of the case making a difference between columns being blocked and everything working as it should! Another confirmation that the Model E is a fragile machine - at least this specimen.

This particular machine was maybe one of the last to be exported from the US to Holland, before shipping became dangerous and expensive. The nameplate is the factory 'generic' in English without any patents.

The rear of the case lists the patents, from the first 1887 to an October 1913 patent.

This Comptometer was used - judging by the wear it was used heavily. The rust on the sides shows where hands have rubbed off the protective lacquer and even the copper coating to expose the steel. The machine was also serviced and repaired in the past. Several replacement screws in the case show it was taken apart completely at least once. The broken keystem had solder remnants; so when that initially broke the machine was still valuable enough to try a repair to keep it in use.

The new Controlled Key Comptometer Model E - repaired! (And fragile.)

Blickensderfer ink idea - more testing

Having found a cheap and plentiful source of very small quantities of oil-based ink, some more experimenting was done. Especially some colored inks; a discount-store had a brand set of colored ballpint pens for a few Euro.

These are very easy to take apart - twist and pull gets the top off and it all comes apart. The nib can be pulled out of the white tube easily (and pushed back too, they re-assemble equally easily).

The ink in these Staedler color-pens is more runny than the BIC Cristal pens, so simply holding the tube over a fresh roller lets the ink drip onto the felt. Three new rollers, with each a different color ink:

The ink is still quite slow to soak into the felt. After an hour it is deeper, but by no means uniform. If the soaking has 'halted', as e.g. on the green one, another drop is added.

With the lids on, these were left overnight in their trays to fully soak through the entire roller. The next day the felt is completely colored, with ink distributed throughout; the purple roller was then placed on the machine:

To make sure the felt is properly saturated through and through, the obvious thing to do is to add another drop of ink just to make sure. That works, but probably also leads to an over-inked roller.

When typing with an over-inked roller too much ink is rolled onto the type. This causes excess ink to be flung off the type at impact with the paper. This causes streaking or splatter around the characters - seen on the left-most 'test' typing. The remedy for bad inking caused by too full a supply of ink is rubbing the pad with blotting paper, as advised by a Blickensderfer instruction manual:

Not blotting paper, but a bit of card works just as well - removes excess ink from the surface. 

This does indeed eliminate streaking and splatter - the right-most 'test' typing. The letters are printed clear and reasonably crisp, but noticeably fainter than before.

The typing quality is definitely not on the level of a typebar machine with ribbon inking. On the other hand the typewheel used here is new, made with a 0.08 mm layering. (The Attic typeface used here is also a bit too ornate for typewriters perhaps.) Earlier experimenting with water-based ink and heavy inking gave much better writing, but were impractical because of soaking through the paper and drying out within the day (or hour!). 

Overall the oil-based ink works, but the printing quality is very sensitive to the level of ink on the roller.

Looking closely at an original typewheel's type ('slug'), the top is not sharp as a typebar machine's type, but has a definite flat surface facing the paper. That flat surface is probably by-design. That surface is there to take the ink and act as the stamp. Ribbon-inked machines that work on impact do not need this flat surface, the impact onto the ribbon material presses ink onto the paper.

The needs of the 'stamping' process will be behind the typing advice given in the Blickensderfer instruction manual. On a regular ribbon-inked machine the typing is 'staccato'; the fingers strike the keys to fling the type against the ribbon and paper at speed, to rebound immediately. The inking is done by impact; near-instanteneous. On the Blickensderfer typing should be not by striking the keys, but rather by pressing them down - the type should be pressed against the paper to act as a stamp. This will be needed to give the ink time to transfer onto the paper, as in stamping. It's explained in the manual: the key must be followed down until the type is stopped by the paper.

Even though pens give low-cost oil-based inks (which oil?) and do work, the ink of these particular colored pens is not great - too runny and relatively light on colorant. Current impression is that the BIC Cristal inks are better suited for the Blickensderfer; thicker, sticky and very dense. Also the type itself on newly-made wheels can perhaps be optimised for the printing process. 

Another realisation is that ribbon-inking is a very good, robust solution for typing.

More tinkering (playing) may be possible on getting this Blickensderfer 5 to type as it would have done 120 years ago :-)