Every number Meridian gives you comes from somewhere. This is where, and it includes the places the research disagrees with us.
How much does hot air cost a runner?
It has been studied properly, in large samples, over decades. The answers are consistent enough to build on and uncertain enough that anyone claiming a single exact number is selling something.
Meridian's model reproduces nine published findings. It matches eight of them. The ninth it disagrees with, on purpose, and that is on this page too.
Seven marathons, up to 36 years of results each, hundreds of thousands of finishers. The effect is larger for slower runners, who are out in it for longer — which is why Meridian scales the penalty with time on feet rather than distance.
Ely, Cheuvront, Roberts & Montain (2007), Medicine & Science in Sports & ExerciseStockholm Marathon finishers across many years show a gradient that steepens as conditions worsen. Doubling the heat does considerably more than double the damage. Meridian's heat term is quadratic for this reason, not because a curve looked better than a line.
Vihma (2010), International Journal of BiometeorologyThe received wisdom was that nothing happens below about 600 m. It does. The decline is linear and it starts almost immediately, which is why Meridian's altitude threshold sits at 300 m rather than the traditional figure.
Wehrlin & Hallén (2006), European Journal of Applied PhysiologyA meta-analysis of 96 studies and 1,048 participants. Ten to fourteen days of regular heat exposure recovers a substantial fraction of what the heat takes. This is why Meridian works out where you actually train instead of asking you how tough you are.
Tyler, Reeve, Hodges & Cheung (2016), Sports MedicineTreadmill work across gradients from −45% to +45% produced the polynomial Meridian uses for elevation. It is one of the few parts of the model that is close to settled physics rather than statistics.
Minetti, Moia, Roi, Susta & Ferretti (2002), Journal of Applied PhysiologyA pooled analysis across endurance running puts the marginal cost near 0.2–0.4% per degree of heat index. Meridian predicts roughly 0.7% for a marathoner, which is close to double.
We think the gap is real and mostly explained: that analysis pooled events far shorter than a marathon, where heat costs much less, so the average is diluted by races our model is not describing. The marathon-specific research points the other way and we have gone with it.
Two peer-reviewed sources genuinely disagree here. We picked a side and wrote down why. If real runner data lands nearer the other one, the duration curve is the first thing we will change.
Mantzios et al. (2022), Frontiers in Physiology — the finding we do not matchNine anchors drawn from that literature run against the model automatically. Any change to any constant has to keep passing them.
| Anchor | What it tests | Result |
|---|---|---|
| Ely marathon slowdown | Cost across a 15°C swing | pass |
| Ely elite marginal | Faster runners suffer less | pass |
| Vihma gradient | Slower runners suffer more | pass |
| Wehrlin 1,500 m | Aerobic loss at altitude | pass |
| Wehrlin 2,500 m | Same, higher up | pass |
| Gore transfer | Capacity lost vs pace lost | pass |
| Tyler acclimation | Adapted vs unadapted at 32°C | pass |
| Minetti gradient | Energy cost of a hill | pass |
| Mantzios per-degree | Marginal cost per degree | disagrees |
Dew point.
Meridian is built on the idea that humidity matters as much as heat — that 75°F air over a 70°F dew point is a different run from 75°F over 45°F. Every coach knows this. Every runner has felt it.
But three large studies looking for dew point as an independent predictor of marathon performance did not find one. The effect kept collapsing into temperature. That is uncomfortable, and pretending otherwise would make this page worthless.
Our position: the studies measured field marathons where temperature and humidity move together, which makes the two nearly impossible to separate. The way to settle it is real runners' own histories across genuinely different conditions. That is the work in front of us, and if it says we are wrong, this page will say so.
A sensitivity analysis over every constant in the model found one that dominates: the threshold where heat starts costing anything. Its influence is about ten times the next largest.
It is a coaching heuristic. It is not measured. If something in Meridian is wrong, it is most likely that, and we would rather you heard it from us.
TestFlight is opening to a small group of runners across a range of climates.