Quote:

Once the main touch points are fixed, the remaining thing to worry about is head position.

In addition to head position, what about sliding forward or backward on the seat? What about scapular retraction or protraction? What about increasing or decreasing natural spinal curvature? Could these things affect aerodynamics? Do you know what the magnitude of these changes would be, for Tom, on a P3C and P2K? I'm just trying to get an idea of the potential error that could be introduced by these changes in position.

Quote:

small changes in head position will affect variance of the estimate, not its location.

You seem to trust that Tom did his best to maintain the same head position on both bikes, and I'm sure he did try, but from your comments above do I correctly interpret that you are ignoring the possibility that Tom consistently held a different head position on the two bikes? You seem to be allowing for the possibility that he moved his head around more on the P3C, but not that he consistently held his head in a different position. Please correct me if I've misinterpreted.

Tom has worked to optimise the repeatability of his CdA, and the data suggests that his position on the P2K in the P2K vs P3C test is similar to his previously reported position. So, as you say, we know he wasn't sitting up on the P2K compared to the position used to produce previously reported data. On the other hand, do we know that this highly repeatable CdA is the very best that Tom could possibly achieve using the contact points that were used for both the P2K and P3C? If it's not, then there would be scope for him to improve his CdA on the P3C.

Quote:

Parameter location is determined by the tilt of the profiles; good precision produces lap profiles that look identical, while poor precision distorts the lap profiles. I don't have the data, only the lap profiles from the plots, but I'd say the precision of the P3C laps is only slightly larger than for the P2K laps.

If he consistently used a different position on the P3C to reduce CdA, would this be obvious from the data?