Is That Autograph Real or Auto-Penned?

The frustrating thing about an autopen is that it deposits real ink on real paper. The fingernail-across-the-Sharpie test passes. The "is this an inkjet print" test passes. From across the room it looks fine. The machine is the reason a lot of NASA fan-mail responses from the 1960s and 1970s, plenty of presidential thank-you notes, and a non-trivial slice of celebrity fan-mail signatures look real on the surface and fail the moment you compare two of them.

How the machine actually works

An autopen holds a real pen — usually a Sharpie Fine, a Pentel Sign Pen, or a ballpoint — and traces a stored path. The path was recorded once, from a real signature the signer made on a digitizing tablet or a templated metal die. Modern machines (Atlantic Plus, Damilic Signascript, the older Autopen 80) can vary speed and pressure slightly so two outputs do not literally overlay. They cannot reproduce the natural variance of human motor control.

The five tells, in the order I check them

1. Two examples overlay

If you can find a second alleged signature from the same signer (same era, same substrate is best but any era will do), scale them to match and overlay. Two human signings will be similar but never identical — height varies by a few percent, the angle of the descender on a "g" or "y" drifts, the connecting stroke between two letters lands a millimeter off. Two autopen outputs from the same template overlay almost perfectly. This is the single highest-confidence test there is. If you have a known autopen pattern in your reference folder — the Buzz Aldrin autopen patterns and the Reagan autopen patterns are well-cataloged in the community — this catches them in seconds.

2. Line width does not change

Hand-signed downstrokes are thicker than upstrokes because the weight of the hand presses harder going down. The crossbar on a "t" is usually thinner than the vertical stroke. Autopen line width is constant, or it varies in a programmed way that does not match human muscle mechanics. On a Sharpie this is harder to read because the marker masks pressure variation; on a Pilot G2 1.0 or a Pentel Sign Pen the effect is obvious.

3. Curves are too clean

Hand-drawn curves carry micro-tremors. Look at the inside of the loop on a "g" descender or the bowl of an "a." A real hand leaves tiny irregularities that come from the tendons in your wrist and the way your pen rolls slightly. An autopen curve is mathematically smooth in a way no human hand produces. Under a 10x loupe this is unmistakable.

4. Pen-down and pen-up are identical

Where the pen first touches the paper and where it lifts off varies in real signing. A signer puts the cap on the photo at slightly different angles each time. The first dot of ink lands a millimeter to one side or the other. An autopen lands the pen in exactly the same spot every time, because the path is deterministic. If five examples from the same signer all start with the entry dot landing in the identical position relative to the first letter, you are looking at a template.

5. No context shows in the line

Real signing carries the signer's mood. A hasty stage-door grab looks different than a careful inscription to a kid. The Sharpie tail is shorter when the signer is rushing. The first letter is bigger when they have time. Autopens have no context. The machine signs at 11:30 a.m. exactly the way it signs at 4:15 p.m.

Who actually uses these

U.S. presidents have used autopens since the Eisenhower administration; every president since has used one for routine correspondence and bill-signing. NASA astronauts — especially the Apollo crews — auto-penned thousands of photos in the 1960s and 1970s because the volume of fan mail was unmanageable. Many current actors, athletes, and politicians use them for fan-mail responses. The Sylvester Stallone secretarial era is famous, but Stallone also has documented autopen patterns from later years that circulate alongside the secretarials — and the two are different problems requiring different evidence to detect.

An autopen-signed presidential thank-you letter is still a historical artifact and the autograph community values it as such. It is not the same as a hand-signed piece and it should not be priced as one. The honesty matters — sellers who knowingly market autopens as in-person hand-signed are the problem, not the machine itself.

What I cannot tell you from a single photo

Some autopens are subtle enough that a single example, viewed in isolation, looks plausible. The detection method requires either a reference pattern from the cataloged autopen library or a second example to overlay against. If I am looking at one photo with no comparison material and the line behavior is ambiguous, I will mark the piece Inconclusive rather than guess. That is the correct answer when the evidence is not there.

For pieces inside a known autopen window — presidential correspondence post-1980, NASA Apollo-era 8x10s, Stallone late-career fan mail — the burden of evidence is higher. I want to see a second exemplar from the same era, ideally with verified provenance, before calling a piece authentic.