Comments on: Autopen

Autopen

azul — Sat, 05 Jun 2004 16:40:33 -0800

Is the autograph real? Dwight D. Eisenhower, John F. Kennedy and of course Ronald Reagan have used it - the Autopen - a machine which signs an autograph in the celebrity's handwriting. It was developed in the early part of this century but only became popular in the late 1940's. Learn more at Stephen Beck's Autograph Pages

By: aladfar

aladfar — Sat, 05 Jun 2004 17:06:36 -0800

Interesting. Any info about how the autopen functioned? How is the original signature "programmed"? Any idea where an old machine might be picked up?

By: dhartung

dhartung — Sat, 05 Jun 2004 22:10:00 -0800

A little more information here (image). The autopen can be thought of as a special-purpose plotter; the analog program, the matrix, was a metal trace of the signature, probably separating the strokes into individual traces beginning from the same base point. The position of the matrix would determine pen-up and pen-down conditions.

By: ALongDecember

ALongDecember — Sat, 05 Jun 2004 22:15:17 -0800

There's a short piece about the Autopen in William Poundstone's 1993 book Biggest Secrets (appears to be out of print). As of that date, here are some specs. They're made by the International Autopen Co. of Sterling, VA. It's about 100 pounds, uses regular power. Two signatures are submitted to the company, and a "matrix" is made by the company that simulates the pen movements. The signatures put out are almost identical, and uses real pens and pencils. Paper must be fed to the machine by hand. Cost is about $3000 dollars.

By: ALongDecember

ALongDecember — Sat, 05 Jun 2004 22:35:11 -0800

This page shows that the Commonwealth of Pennsylvania (we're not a state!) bought an Autopen in 1997 for $3,180. This shows the RNC buying one in 2004. $3,598.

By: tss

tss — Sat, 05 Jun 2004 22:43:37 -0800

Interesting---the edge of the crescent-shaped "matrix" object in dhartung's link is wavy like a key. Clearly that's the signature pen position data. I don't see any other obvious data on the matrix, so it looks like they only need two dimensions to describe the signature, plus pen up/down information. Makes sense given that the signature is a 2-D curve. I never took kinematics, so I can't say much about the linkage seen in the picture of the machine. Given that it seems fixed in place, and given that the two white wheels seem well-placed to change the angle and distance (by changing the angle in the bent central arm) of the pen, the data on the matrix may be the pen location in radial coordinates. The only problem with that is that the range of the angle data is larger than the range of the distance data, since signatures are wider than they are tall (but not too much!). There doesn't seem to be a commensurate difference in the range of the matrix data. We might then expect the machine to be sloppier with e.g. the width of loops than the height. This could be solved in many ways, though---perhaps the angle data is simply deviation from a monotonic progression from the start angle to the end angle, though this will not work well if the signature involves going back and crossing the t's. If it is using radial coordinates, we might expect the machine to be less accurate at the left and right extremes of the signature. Has anyone ever seen one of these work?

By: Down10

Down10 — Sun, 06 Jun 2004 00:24:34 -0800

...developed in the early part of this century... I think you mean last century. ;)

By: davidmsc

davidmsc — Sun, 06 Jun 2004 03:56:09 -0800

I've seen autopens used in the military by 4-star generals (or, to be precise, their secretaries and executive aides), who must affix their signature to many, many things.

By: jfuller

jfuller — Sun, 06 Jun 2004 06:33:35 -0800

> I think you mean last century. ;) It ain't the twenty-first century until I get my personal jetpak.

By: dogwelder

dogwelder — Sun, 06 Jun 2004 07:35:20 -0800

I know someone who used to do Kenny G's autograph for him.