-
Trailing edge, surely?
Trailing edge is for marginal gains, all the hard work is done by the leading edge. A circular cross section so fucks things up that there's almost nothing you can do apart from starting again with something wider that's a good shape, that's why the best aero rims are now wider than the tyres in front of them. It's also why those 90s "Aero" tubes which grafted a trailing edge onto a round tube ended up working better when turned around, and it's why a proper Kamm tail has so little effect on total drag.
The front half (i.e. from leading edge to maximum thickness) of a minimum drag aerofoil is rounded, but it's nothing like a semicircle.
gbj_tester-
I'm not sure that's correct. My understanding, admittedly based on motorsports wind tunnel testing and CFD work rather than for bicycles, is that it's the trailing edge which is most important, hence why a teardrop shape has a much lower Cd figure than a reverse teardrop shape:
Also, if you had the pointed end at the front then the drag profile would be very susceptible to yaw. It might be aerodynamic if the airflow was from directly ahead at all times, but as soon as the airflow started coming at it from an angle you'd get horrible vortices off the sharp leading edge.
I think the Cervelo RCA white paper shows it most clearly at Figure 16. If the leading edge was most important, then the oval shape would be superior to the teardrop aerofoil shape.
Edit: Found figure 16 as a .gif, it's here:
P.S. It also shows that a Kamm tail (the third profile) is better than a circle, but still nowhere near as good as a teardrop shape with the same front area.
Brommers
Trailing edge, surely? A teardrop shape is as aerodynamically efficient as it gets, with a rounded front end and a pointy tail. If you look at any fast road or TT bike which has had wind tunnel time, it'll have a rounded front edge on the HT and then a tapering section behind it.
The majority of the drag doesn't come from the air hitting the rounded front of the head tube - it comes from the vortices from the trailing side of a round tube which are eliminated (or at least substantially reduced) by a tapered trailing edge.