Perhaps it would also have looked more linear if the individual Hb responses to infusions would be taken into account (some guys develop haemolysis after infusion) and not lumped together, we don't know how much each individual Hb increased. But regardless of how one shows the data, it is really difficult to circumvent the conclusion that there was practically no performance boost after the c:a 6 % increase in hemoglobin after the 235 ml. Even the authors write that "the transfusion was a mix of replacing the lost cells and adding additional" and also elsewhere that "[i]t may be speculated that the initial normalization of the blood volume is contributing the most to the performance enhancing effect while the additional increase in blood volume contributes less".
Here is the individual data recalculated from the FIGURE 2. There might be a watt missing from here-and-there, but the average improvements are exactly 4.7 % and 0.9 % as stated in the paper.
291 307 310
217 223 220
202 215 209
203 210 219
199 207 208
196 204 208
186 199 205
From this material, it looks that nobody completely tanked after the 235 ml infusion (worst c:a -2.8 %) and unlikely what one might've expected from the protocol, only one participant responded more to the 235 ml RBC infusion than to the lower one (c:a +4.3 % vs +3.4 %) and even when he was the highest responder, he didn't even reach the average boost of the 135 ml group. There is no way never to prove it, but the data would be far more consistent and in the line with the other blood doping research if there blinding wasn't that good in the first test and the 4.7 % is a drastic overstatement of the "real" physiological improvement.
But - as Dr. Coggan pointed out - this is a small cohort and it causes now-and-then certain anomalies into the data.
Aragon, why do you link polycythemia to power? Is that all you think about? You seem pretty obsessed with this idea.