Observations

Synopsis

This instrument offers the possibility of a tenfold increase in resolution over a 'traditional' spiral design. Given that, it might seem surprising that the calculator didn't become a major contender in the marketplace.

There are some issues with the logarithmic spiral technique that probably were a factor.

The spiral

The main limitation of this Log Spiral is inherent in the logarithmic spiral itself.

The inner turns are much closer together than the outer turns, and this disparity increases with the number of turns.

As with other spiral scales to increase the scale resolution, the number of turns must increase. Also, you need a bigger Log Spiral than an Archimedean spiral to have the same overall scale length, or conversely, if two spirals have the same diameter and the same number of turns, the traditional spiral will be longer. In the cases I've tried it looks like the Vernier capability more than compensates for that difference.

At any rate, what this means is we need to make the calculator larger we would need with a traditional spiral design, to avoid excessively small separation between the first turns. I have a 25 turn version below side by side with a Gilson Binary that shows this.

Vernier limitations

It should be apparent that the Spiral vernier is harder to use near the centre of the spiral. They begin to curl inward, and are closer together in the centre, and also intersecting the Spiral at a sharper angle towards the center. Coupled with the fact that the turns are closer as well, it's basically tough to take a reading.

George Macdonald was completely aware of these problems and attempted to address it in his patent.

He proposed several mechanisms to keep the instrument readable, including:

  1. Removing leading digits from the numbers on the disk. Macdonald added an extra scale on the Vernier cursor that allowed you to infer the leading digits.

  2. Writing the numbers horizontally (perpendicular to the line of the spiral, rather than parallel with it)

  3. Replacing tick marks with dots.

  4. Extending the Vernier to the opposite side of the cursor in large turn spirals, which allows for even and odd coil numbers to be separated. He also doubled (halved?) the log scale to fit 2 cycles on the outer circle.

Manufacturing costs

I reckon that this would be an expensive calculator, particularly if you were chasing the high-precision market. The framed vernier and index that Macdonald describes includes 3 locking mechanisms, various techniques to allow for maintaining precise alignment, and for the really long scale versions, mechanisms to provide for fine adjustments.

Comparison between a 25 turn Gilson Atlas and the Macdonald Log Spiral set to 25 turns

The following shows how the invention might have stacked up with the popular Gilson. In order to make the letters readable, I had to raise the size of the log spiral scale to 9.5 inches in diameter, compared with the 8 inches needed for the Gilson.

9.5 inches happens to be the same dimension that Macdonald suggested for a 50 turn log Spiral; I haven't been able to produce a readable version of a 50 turn model at that size. It apparently needs somewhere around a 15 inch diameter disk.

On the other hand, I did not try any of the the mechanisms mentioned above.

Macdonald Log Spiral

This Log Spiral has a 9.5 inch outer diameter. The scale length is 614 cm, about 24 feet. With a Vernier, 5+ digits are feasible. This one I have not built, but the the spacing between turns at the centre is about 1.2 mm, which constrains character sizes to about 3 points.

Compare this scale with the Gilson Atlas below:

Gilson Atlas

The Atlas has a scale length of about 11 meters or 35 ft, and provided 4 digits resolution.

The turns are separate by a uniform 2.5 mm spacing. The letters were about 1/2 that, close to 1.2 mm, or about 4 points.

My apologies for substituting a replica; I don't have a photo of a Gilson Atlas. My dimensions are taken from C. Hamann's page on the Gilson Atlas.

Another Logarithmic Spiral calculator

I have seen one other patent (US268221 Calculator for Miller Use - by Haight and Segur) that uses a logarithmic spiral rather than an Archimedian spiral.

If I find any others, I'll add them here. I may attempt to replicate some or all of them.