Disclaimer: I am learning all of this in German, and I have German books. So I might be wrong in translating some of the technical terms I use. I may try to buy the manuals in English one day and correct my translations.
My 5th day of lessons (we call it palying) with Erich Hirsch was only half a day, but we used every minute of it. (As I live in Sweden, it’s not easy for me to get to Hamburg).
This day we concentrated on the keyboard. We set the different widths by turning the lever on the paper tower and by keeping the (<-) key pressed all the time. In the process, we drove the carriage to the end stop a number of times.
We looked closely at the interaction of the upper and lower keyboards, why there are 31 holes and 33 crossbars.
31 holes: 14 rows, 14 columns, coarse adjustment, fine adjustment and one hole for S. O15, the quad, is controlled without holes.
33 crossbars: O and 15 also have bars and punch bars, but no punches.
The first hole is S and for a variable, S + 2 is always punched. I now know how to easily distinguish the wide keyboard from the narrow one. (The wide one has 12 columns, the narrow one 11, and this is best seen in the row of numbers). And that the translation meachnics of the wider keyboard have narrower bars.
We then took a closer look at how the information from the rows is transferred to the scale via the unit insert. In a typical matrix frame, the first row has 5 units and the last 18. In the unit insert, the smallest value is 4, which is punched as the minimum width for each variable. We looked at various unit inserts, including those where Erich had made new combinations from existing inserts. We looked at a case where the rows do not have continuously increasing unit values (e.g. 14-15-14-15-16). This may be necessary when moving units to make full use of the frame. In combined rows (i.e. where several rows of matrices have the same unit value) the bars are lifted by springs and not by the mechanism, which would otherwise block.
The cam disc on the paper tower establishes the exact timing between punching and paper transport.
Finally, Erich gave me another table as a practice, this time with decimal values, which should be aligned with the decimal point. Set the widest column and then move the pointer further for the narrower columns using the coarse justification keys (for this, you have to know which key punches the corresponding unit values, and that then corresponds to the unit value of the corresponding row in the matrix frame). Each column is simply justified, unlike the row. So the column is first justified coarsly, then finely. The row is first struck coarse, then coarse+fine so that the row expulsion mechanism is triggered.
The number column is set purely with fixed blanks, no variables, counting all digits as half-quads (9 units). The column should be 6 cicero wide, which corresponds to 6½-6 (i.e. 6½ quads and 6 units) at 11¼ set of the font we have chosen. The number “647.30” is 2½-5 wide, then no matter what font, as long as they are table digits. So you strike 2 quads each before and after, and the missing one unit is ignored.
If you need 2 units, you can take 5, then 6, which gives you a half-quad+ 2 units.
And it worked. I forgot to double justify before the first line, so that it would be the last line when casting. I also forgot to hit an extra quad after the first line, so that the casting machine would stop. Erich, being an experienced caster, noticed this immediately. He corrected it on the fly.