the linear-intercept method
/ Heyn -> HYNE /
How do you measure grain size without tracing every wiggly grain by hand? The linear-intercept method uses a wonderfully simple trick: draw one or more straight test lines of known total length across the micrograph and count how many grain boundaries they cross. Grains cut by a longer average slice are bigger; grains cut by a shorter average slice are smaller. It is like estimating the size of paving stones by walking a tape measure across a courtyard and counting the cracks you step over.
The recipe: lay test lines of total true length L on the image (corrected for the magnification), count the number P of grain boundaries the lines intersect, and compute the mean linear intercept, l = L / P. That l — the average distance a random line travels within one grain — is the reported grain-size measure; it can be converted to an ASTM number if wanted. Example: test lines totalling 500 micrometres of real length cross 20 boundaries, so l = 500/20 = 25 micrometres. Using many lines in several directions averages out the irregular grain shapes and any elongation.
The method is fast, needs no software, and is enshrined in ASTM E112 alongside grain counting. It is also good statistics: counting more intersections tightens the estimate, and using lines at several angles reveals whether grains are elongated (the intercept then depends on direction — a fingerprint of a deformed or directionally solidified structure). Honest caveat: the intercept l is not the physical grain diameter — it is a mean chord through randomly sectioned grains and runs smaller than the true 3D grain size; a stereological factor is needed to relate the two.
A circular test line 250 micrometres long (true length) is superimposed on three different fields and crosses 30, 28 and 32 grain boundaries, 90 in total over 750 micrometres. The mean linear intercept is l = 750/90 = 8.3 micrometres — a fine-grained material.
Count boundary crossings, divide total line length by the count — no grain tracing required.
The linear intercept l measures a mean chord, not a grain diameter, and it is systematically smaller than the true mean grain size; convert with a stereological factor if you need the real dimension. Count enough intersections (roughly 50 or more) for a trustworthy average.