Steel Carpenter's Square

Write down the question this blueprint answers: how do you check a square when the only thing you have to check it against is another square?

Choose the straightest edge of the wooden board. Sand it flat with 220 grit until a steel rule laid along it shows no light underneath.

Hold the try square against that edge. Scribe a line across the board with the marker.

Turn the square over on the same edge, register it on the same point, and scribe a second line on top of the first.

If the two lines diverge, the square is out. The gap you see is twice the error, because flipping adds the fault to itself. Halve it before you record it.

Measure the gap 200 mm from the pivot point. Divide by two, then by 200: that is the error per millimetre of blade.

Run the same flip test on the large framing square. Long blades magnify a small angular fault, so expect a larger gap for the same error.

Stand the machinist square on the surface plate beside the try square blade. Hold both to the light. The line of light between them is the difference.

Slide feeler leaves into that light gap until one just holds. Record the leaf thickness — now the error is a number, not an impression.

Cut three plywood strips at 300, 400 and 500 mm. Any triangle in that ratio is right-angled — Hawes' carpenters knew this centuries before they could buy a steel square.

Glue the three strips into a triangle. When dry, flip-test it exactly as in steps 3-5.

Put it in a bathroom or outdoors overnight. Re-run the flip test in the morning and record the change. This is the failure Hawes' steel square removed.

Write the four squares in order of measured error, smallest first. Keep the best one for layout and use the rest for rough work.

Silas Hawes (1779-1854), blacksmith, South Shaftsbury, Vermont. United States patent granted 15 December 1819 for a steel carpenter's square.

Why there is no patent number here. On 15 December 1836 — seventeen years to the day after Hawes' grant — fire destroyed the Patent Office building in Washington and with it every original file. About 10,000 patents burned; roughly 2,845 were later reconstructed from inventors' own copies and are now called the X-patents. Hawes' was not among them. His claim text, drawing and number are gone, and honest sources cannot supply what was destroyed. Where batch 35's blueprints could quote a sheet, this one cannot, and it says so rather than inventing a plausible number.

What is documented. Hawes was making squares in quantity in South Shaftsbury in the 1820s; the town became a centre of square-making, and the trade later concentrated in Shaftsbury and then in Connecticut. The saw-blade origin story is traditional and repeated in local histories; treat it as tradition, not as evidence.

The real content of the invention. A one-piece steel square has no joint, so it cannot rack; it does not swell or shrink with humidity as a wooden square does; and because it was rolled and then stamped, every square off the line was the same square. That last property is the one that matters industrially — not accuracy in itself, but repeatable accuracy you can buy. Every layout in the shop starts from it.

The flip test is older than the patent and still the standard. It works because it needs no reference square at all: it compares the tool against its own mirror image. Any error appears doubled, which is also why it can detect faults far below what you could see by eye. Machinists still check squares this way.