Pratt, Howe or Fink — How to Pick a Roof Truss Shape

A roof truss's name — Pratt, Howe, Fink — isn't a style label. It describes which way the diagonal members lean, and that one choice decides whether those diagonals carry tension or compression under the roof's own weight. Get it backwards and you've sized half the web members for the wrong kind of force.

The one thing that actually differs between them

Take a simple triangulated truss on two supports. In every panel, a diagonal connects the top chord to the bottom chord. It can lean either way, and under gravity load, that direction decides everything:

• Pratt: diagonals lean in from each support toward mid-span. Under the roof's own weight, they're pulled straight — tension. The short vertical members take the compression instead.
• Howe: the mirror image. Diagonals lean out from mid-span toward each support, so under gravity they're squeezed — compression. The verticals carry tension instead.

That's the whole difference. Same triangulated pattern, same number of members, opposite diagonal direction.

Why Pratt is the usual default in steel

Tension members don't buckle. A slender angle carrying tension is just as effective as a stocky one, so a Pratt truss lets you keep the diagonals — usually the longest members in the panel — light. A Howe truss puts those same long diagonals into compression, where slenderness limits bite, so they end up needing a heavier section for the same load. That's why Pratt is the more common default for an all-steel truss, and why it's the tool's starting option.

Howe still has its place — it was the standard for timber trusses, since compression suits timber's grain better than tension does, and you'll still see it specified to match an existing drawing or a particular fabrication preference. It isn't wrong, just usually not the lighter option in steel.

Fink is a different kind of choice

Fink (also called a W-truss) isn't a mirror-image variant — it's a fixed pattern built for pitched roofs specifically. Instead of one diagonal per panel, each half of the truss gets a short W-shaped arrangement that keeps every member short near mid-span, right where a plain Pratt diagonal would otherwise run longest. That's why Fink is the shape most people picture for an ordinary pitched roof: it's naturally efficient for that geometry. The trade-off is that it's a fixed four-panel layout, not something you stretch or subdivide further — for a longer span or a different panel count, you're back to Pratt or Howe.

The twist: wind doesn't know which truss you drew

All of the above describes what happens under the roof's own weight. A lightly sheeted roof in a windy zone tells a different story — uplift can exceed the dead load enough to reverse the net force in a member entirely. A Pratt diagonal that's comfortably in tension under dead and live load can find itself in compression once wind lift dominates a combination like 0.9 DL + 1.5 WL, and by then it's a different check altogether: slenderness and buckling, not just tension capacity.

This is exactly why a truss should be checked against the full set of load combinations — both wind directions, internal pressure both ways, dead load with and without live load — rather than just the gravity case a truss's "name" describes. The member that governs is often not the one you'd guess from the diagram alone.