How do I check my spirit level is accurate?

A level is the only tool that can be confidently, invisibly wrong across an entire job. Nothing about a faulty level looks faulty — the bubble sits between the lines, you fix the shelf, and it slopes. The test takes half a minute, needs no reference surface at all, and should be done on any level that has been dropped or that has been in a van for a few years.

What you'll need

• The level under test
• Any reasonably flat surface — it does not need to be level
• A pencil

Step by step

1. Find any flat surface.A worktop, a windowsill, a plank on the floor. It does not have to be level, which is the elegant part of the test — you are comparing the level with itself, not with a reference.
2. Set the level down and mark its position.Pencil lines at both ends. The level has to go back in exactly the same place, because moving it changes what it is reading.
3. Note the bubble position precisely.Not just 'roughly centre' — note which side of centre it sits and by how much relative to the vial's graduation lines. If it happens to read dead level, that is fine too.
4. Turn it end for end and put it back in the marks.Rotate it 180 degrees about a vertical axis so the end that was on the left is now on the right, keeping the same face upwards. Set it back precisely between your two pencil marks.
5. Read it again and compare.If the bubble sits in the same place relative to the vial, the level is accurate and the surface is simply not level. If the bubble has jumped to the other side by a similar amount, the level itself is out.
6. Interpret the size of the jump.The bubble moves twice the level's error, because reversing adds the error to itself. So a level that reads one graduation right, then one graduation left, is out by half a graduation. That doubling is what makes a tiny fault visible.
7. Test the plumb vial the same way.Hold the level against a door lining, mark its position, note the bubble. Then rotate it half a turn about its long axis so the opposite face is against the lining, put it back in the marks and read again. Same rule applies.
8. Repeat for each vial and each face.A three-vial level has a horizontal, a vertical and often a 45. And many levels have vials readable from both sides — check the face you actually use.
9. Decide what to do about a failure.Some levels have adjustable vials, held by a small screw or an accessible plastic housing; those can be corrected by adjusting out half the observed error and re-testing. Most sealed levels cannot be adjusted. A level that is meaningfully out and cannot be adjusted should be thrown away, not kept for rough work — it will get used for something that matters.

Tips

• Do this to a new level in the shop or on the day it arrives, while it can still be returned. Cheap levels fail more often than people expect.
• The same reversal principle checks a laser level: set it up, mark the line on a wall, rotate the unit 180 degrees on its base, and see whether the line lands in the same place.
• A level dropped on its end is the classic way to knock a vial out. Test it immediately rather than at the end of the job.
• Keep the level clean along its bearing edge. Dried plaster, paint or adhesive on the underside tilts the whole tool and no vial test will reveal it.
• Longer levels are more accurate over a run for a reason: the same angular error produces the same slope, but you need fewer repositions to cross the room.

Common mistakes

• Testing on a surface you believe to be level — It confuses two questions. The reversal test deliberately does not care what the surface is doing, which is why it works anywhere.
• Flipping the level over onto its other face instead of end for end — That tests a different thing — the parallelism of the two faces. Both tests are useful, but the end-for-end reversal is the one that finds a vial error.
• Keeping an inaccurate level 'for rough work' — It will be picked up by someone, for something that matters, on a day when nobody remembers. Levels are cheap and the consequences of a wrong one are not.

Questions people ask