# Checking squareness of a T-square

There are well known ways to self-prove a straight edge, combination square, or machinist square, to the accuracy needed in woodworking. Those methods only need the tool itself, a pencil or knife, a flat surface (for straight edges) or a flat surface with a true edge (for the squares.) They also amplify the error by twice, which helps the human eye spot it.

Is there a comparable way to check the squareness of a "T square"-like square? I'm thinking of things with a stock/fence that is below the plane of the blade/beam, such as an architect's drafting T-square, a drywall square, a shop made circular saw cross cut guide or router dado guide, etc.

You cannot flip them over like a combination square because then the fence is not registering against anything. Referencing against the opposite edge of a piece of wood just reproduces the same angle, not a flipped angle; even if it didn't, you are then trusting the two reference edges are parallel.

I am aware of the 3-4-5 method, but I don't feel like I can read a ruler or measuring tape to the mere thousandths that the straight edge/combo square truing methods afford. You can compare to another square, but often the square I want to test is significantly larger than my most trusted 12" square.

Is there a clever method I am not aware of?

• You can use a T-square flipped over if you tilt it slightly so one side still registers against the edge of the surface. I've never tried checking squareness that way, though. Commented Jan 31, 2020 at 22:01

I am aware of the 3-4-5 method, but I don't feel like I can read a ruler or measuring tape to the mere thousandths that the straight edge/combo square truing methods afford.

This method can in fact be perfectly accurate and was what I was going to suggest primarily, as it's one standard way of laying out for a self-built T-square, or anything where you need a large 90° drawn from scratch with accuracy1.

I wouldn't use a tape for this, too much wiggle room and the tab will be a hindrance, but of course you would need a 1m or yard rule instead which not everyone has or wants to get2.

Is there a clever method I am not aware of?

Add a lip to the straight edge you reference off so you can register the fence against the edge once the square is flipped over.

Technically add two pieces, with enough space between them to easily pass the blade or beam. Although it should be doable for any woodworker to plane a scrap of wood down to a uniform thickness there's no reason not to use existing material if it's on hand and fits the bill. I'd go with tempered hardboard or MDF as, barring edge damage, these should be absolutely consistent (but confirm). The pieces can simply be glued in place with superglue, but just double-sided tape might actually be sufficient and not introduce any unevenness.

There is another method that may work, but no guarantees. That is to use the factory corner of a sheet of ply, MDF or particleboard. These should be dead square just as the factory edges should be reliably straight, hence their commonly suggested use as the straight edge against which we test try squares etc. Again, you can't assume it's a perfect 90 so you need to check it, but how.....?3 :-)

1 Because I didn't have a reference square at the time I made my first wooden try squares this is the method I used. They're dead-nuts square as far as a fine pencil line can show.

2 Although I personally find mine indispensable. Although not a frequently used tool the few time I've needed it nothing else would do.

3 Back to the 3-4-5 rule again.

• Thanks for your answer. Since you brought up the 3-4-5 approach, are you doing anything special to read the rule with additional accuracy? Or is it a matter of getting a rule that doesn't have as much potential for ambiguity due to parallax, like a transparent rule with markings on the bottom, or one where the divisions are brought down the edge? Commented Feb 2, 2020 at 4:51
• Just being very careful to read it as accurately as possible — looking very close, viewpoint directly over the rule, very thin marks (I was using a mechanical pencil, so thin marks were assured) and rechecking at least once. If you wanted to test the method to see if it's accurate enough for your needs, by working in the centre of the long edge of a full sheet of board material you can check by laying out to both sides of the right angle, sort of like checking a drawer for square by measuring for equal diagonals. Commented Feb 2, 2020 at 8:05
• @ScottHilbert depending on how big your T-Square is, you can use 6-8-10 or 12-16-20, as on the longer distances, smaller errors will be more obvious. i.e. if your square is a very small amount off, it may not be apparent over the 5" diagonal, but may become much more apparent over the 20" diagonal. Commented Feb 3, 2020 at 16:19
• @FreeMan appreciate the note in case others hit this page and do not know you can scale the ratio. Also maybe worth including here is that since drywall squares often have very uneven leg lengths you could use other pythagorean triples like (5, 12, 13), (8, 15, 17) or (7, 24, 25) Commented Feb 3, 2020 at 21:59
• Sadly, @Graphus, many, especially here in the US, aren't taught to think, just to regurgitate for standardized tests. Extrapolation is a skill that is rapidly diminishing. (No offense intended or implied at our OP, just a sad statement of fact.) Commented Feb 4, 2020 at 12:11

If you happen to have a piece of plexiglass make a small edge cut with a thin razor knife. Place the t square on the edge and slide to the razor blade then make a line or scribe a line the 48” length. Flip it to the other side registering the same cut line razor blade, scribe or mark a line and simply compare.

• Hi, welcome to StackExchange. Did you read beyond the title? This is basically the tried-and-true technique the OP themselves refer to in the first paragraph, and then go on to say it can't be used in this context"I'm thinking of things with a stock/fence that is below the plane of the blade/beam.... You cannot flip them over like a combination square because then the fence is not registering against anything." This was the reason for the Q in the first place. Commented Sep 6, 2021 at 7:22