Measuring 8 elite skaters in 1,500 m race simulations corrected the power-balance model: anaerobic power does not follow an "all-out" pattern, aerobic power rises faster than assumed, and skating position changes through the race (knee angle opens, trunk angle decreases)
Participants
n=8 elite speed skaters.
Protocol
Competitive race imitations and submaximal efficiency tests.
External power from movement analysis, aerodynamic models and measured ice friction; aerobic power from breath-by-breath VO2 and measured skating efficiency; anaerobic power as external minus aerobic power.
Key Findings
Gross skating efficiency: 15.8% (SD 1.8%).
1,500 m anaerobic power followed a first-order decay: P_an = 88 + 556·e^(−0.0494t) W.
Aerobic power rose with rate constant 0.153 s⁻¹ after an 8.7 s delay to a peak of 234 W — similar to other competitive imitations but faster-rising.
Anaerobic power did not fit an "all-out" pattern with near-zero values late in the race, as the group's earlier model assumed.
Skating position changed during the event: pre-extension knee angle increased and trunk angle decreased — different from the earlier model's assumptions.
Authors' conclusion: these measurements allow better predictions and show the value of observations in unique subjects for model building.
Practical Application
In 8 elite skaters racing 1,500 m, gross skating efficiency was 15.8%, aerobic power rose quickly (8.7 s delay, peak 234 W) and anaerobic power decayed but did not run to zero as an "all-out" start would, while knee angle opened up through the race. Inference, not tested: pace the 1,500 m as a fast start that leaves anaerobic power in reserve, not an all-out effort from the gun.
Inference, not tested: the rising knee angle late in the race fits the blood-flow cost of the low position (REF-667) — holding position deep into the race is a trainable target.
Not from this paper: how the aerobic/anaerobic mix shifts across other distances is not measured here.
Limitations
n=8, one distance (1,500 m), race imitations rather than competition.
Model-based partitioning: anaerobic power is computed as a remainder, so errors in efficiency or friction estimates carry into it.
How this entry was checked and changed
Content correction 2026-09-27
"Anaerobic dominates 500 m sprints, aerobic share rises with distance" -> not in the abstract; the study measured only the 1,500 m. Removed; title and lead rewritten to the abstract's findings (efficiency 15.8%, P_an and P_aer kinetics, non-all-out anaerobic pattern, position drift).
"Use fast-start-then-hold pacing where the model supports it" -> restated from the measured 1,500 m pattern; distance-general energy-mix advice removed.
Type "model-validation" -> model evaluation; added n=8 and the protocol. Applicability Direct -> Indirect.