Flatness
The control most often written on a mating face, and the one most often confused with parallelism.
What it controls
Flatness limits how far an entire surface may depart from a perfect plane. It is a form control, so it is measured on the surface alone, with the part free to sit however it likes.
That independence is the whole point. Flatness says nothing about where the surface sits or which way it faces — only that it is flat. A plate can be perfectly flat and still meet its mating face at an angle.
The tolerance zone
The space between two parallel planes a set distance apart. The entire surface has to lie somewhere between them.
Try it in 3D
Interactive 3D needs WebGL. The tolerance zone is described in full above.
Within tolerance
Drag to exaggerate the deviation. The part is shown far out of scale so the zone stays visible.
Reading the callout
The whole of this surface has to lie between two parallel planes 0.05 mm apart.
When to reach for it
- A gasket or sealing face, where a gap anywhere on the surface leaks.
- The base of a fixture or a mould plate that has to bed down without rocking.
- A face that will be used as a datum later, which has to be trustworthy before anything is measured from it.
How it is measured
- A dial indicator swept across the surface with the part on a surface plate. This reads the surface relative to how the part happens to seat and picks up the opposite face too, so the total swing is an upper bound on flatness rather than the flatness itself — levelling the part on three adjustable supports gets much closer.
- A CMM taking a grid of points, with the software fitting the minimum separation of two parallel planes that contains them all.
- An optical flat with monochromatic light, for lapped surfaces where the tolerance is well below a micrometre.
Getting it right
- Reach for parallelism or perpendicularity when the surface has to relate to another face — those controls limit flatness as a by-product, so writing both is usually redundant.
- Relate the flatness value to the function, and be explicit about whether size and form are linked. Under ISO 8015 the independency principle applies by default, so a thickness tolerance does not limit flatness on its own — the envelope requirement Ⓔ has to be stated on the size dimension for that link to exist. On an ASME drawing Rule #1 makes the link automatic, which is one of the few genuine differences between the two systems. One exception is worth knowing on a plate: Rule #1 does not apply to commercial stock — bar, sheet, plate, tube — left in its as-furnished condition, so there a flatness callout is the only thing controlling those surfaces on either system.
- State on the drawing whether the part is to be measured restrained, because a thin plate reads differently clamped than resting free. Left unsaid it will be measured free, and buyer and supplier each read the requirement their own way until parts exist. ISO 10579-NR is the note that makes restraint part of the specification.
Common questions
Does flatness need a datum?
No. Flatness is a form tolerance and is measured on the surface by itself. The part is free to be oriented any way at all during the measurement — only the shape of the surface is being judged.
Flatness or parallelism — which should I use?
Use flatness when the surface has to be flat in its own right, such as a sealing face. Use parallelism when it has to be flat and also run parallel to another face, because parallelism against a datum already limits flatness to the same value. Writing parallelism 0.05 and flatness 0.05 on the same face makes the flatness callout redundant.
How is flatness actually measured?
On the shop floor a dial indicator is swept over the surface with the part on a surface plate, usually on three adjustable supports so the seating can be levelled out; the total indicator movement is an upper bound on flatness. On a CMM a grid of points is taken and the software fits the smallest possible separation between two parallel planes that still contains every point, with the fit free to orient itself. That minimum-zone fit is the definition the standard uses, and it is why the surface-plate number is usually the pessimistic one.
Why is flatness hard to hold on thin parts?
Because the part moves. Clamping deflects it, cutting releases residual stress, and heat treatment moves it again after machining. On thin plates the practical answer is usually stress relieving between roughing and finishing, and grinding or lapping last. That is a sequencing question as much as a machining one, and it is worth raising while the drawing is still being written.
These pages explain the published standards for reference and describe the tolerance zones they define. The values shown in the examples are illustrative. What a given part can hold depends on its geometry, material and process, and is agreed per drawing.