Replicas made from patents
In this section, I'll keep my records of investigations into slide rule patents. I build working models from interesting-looking patents, just to see the proposed design looks in real life.
In this section, I have put my hexadecimal slide rule, and a very interesting Logarithmic Spiral device invented by GM Macdonald, others to come.
I intend to use this section to hold 'replicas' (or more accurately, 'working models') for instruments that do not appear to have been manufactured, although I may deviate from that plan in the future. My Nystrom replica was also built mostly from the patent.
The Nystrom instrument was manufactured, so I was able to validate a lot of details using images of those and a published manual for it. The Nystrom pages are in the Replica SR folder.
I have always been interested in spiral models because they provide the easiest way to construct high-precision SR scales (at home). Based on my review of patents, it's clear that inventors also liked the spiral models very much for the same reasons.
The biggest problem with spiral models is that the answer to any computation may appear on any one of the turns of the spiral. There needs to be a mechanism to find the right turn.
The Gilson Atlas was one of the more popular spiral-based designs. It has a spiral scale and a single-cycle log scale on the circumference. It had a special (to me, complicated) mechanism to allow the user to compute which turn of the spiral the answer would be found.
However, I think many people probably used a simple technique: do the computation twice. For a chain calculation, if you first do the computation on the outer C scale, that gives an answer to 3 digits, enough to spot the turn the full-precision answer is on. Then the same calculation, done on the spiral scale, gives a full 4-5 digit result.
Our canny predecessors seem to have agreed that having to do the same computation twice was tedious and, I suppose, not in the true spirit of using a slide rule...you want the answer in one setting.
Contents
I've located over half a dozen patents trying to address the "turn finding" design problem. I'll work through some of them, in more or less chronological order.
1892 - the GM Andrews Calculating Instrument, so far the earliest I've found, has a turn counter on the face, and a scale on the side of the cursors.
1896 - the Joseph Michaelson Computing Instrument, has an automated mechanism to move a pointer to the turn where the result of computation lies.
1912 - Lilley's improved Calculator is quite similar to the GM Andrews model, except it doesn't have a separate mechanical turn keeper and doesn't have moving indices on the cursors. I haven't yet made a working model for this one because it may have been manufactured.
1947 - The Macdonald Logarithmic Spiral calculator. I skipped ahead a few decades because this instrument is so unusual. It does feature turn counting but that logarithmic spiral is the real show for this one.
1969 - Hexadecimal Calculator. This was formerly in my 'replicas' folder; years ago I made a working model from a patent written by a team of people. The same team additionally obtained a patent for an Octal (base-8) Calculator that I might someday attempt.