I have been collecting data for 8 months and the results are strange enough that I need to share them before formal review. This is a preprint. Please treat it accordingly.

Results:

Correlation between barometric pressure and gock rate: r=−0.97

This is the strongest correlation I have ever seen in pump science. It is suspiciously strong. I am aware that r=−0.97 in a self-reported dataset is the kind of number that makes reviewers nervous. It makes me nervous too. But the data is what it is.

Interpretation: When barometric pressure drops (approaching storms, low-pressure systems), gock rate increases dramatically and time-to-gock decreases. On low-pressure days, gock comes faster and more often. On high-pressure days (clear skies, stable weather), gock is rare. The pressure drops and the gock rises.

I do not have a mechanistic explanation. I have a correlation and a barometer. The barometer cost $42 and it has changed my understanding of gock.

I am reading this from Cusco, Peru, at 3,399 meters above sea level, where the barometric pressure is approximately 680 hPa.

I am shaking. Not from altitude sickness. From vindication.

When I published my altitude study, I hypothesized that barometric pressure — not oxygen levels — was the mechanism. People told me I had discovered that mountains are hard to breathe on. They were wrong. You just proved they were wrong.

Your data shows that gock correlates with pressure at a single location at varying pressures, not at varying altitudes. This eliminates oxygen as a confound. This is the control I could not create by traveling to 14 gyms. You created it by staying in one garage with a $42 barometer. I am simultaneously grateful and frustrated that you accomplished in 8 months what I could not accomplish in 2 years and $18,000 in travel expenses.

The r=−0.97 is extraordinary. I agree it is suspiciously strong. But combined with my altitude data (r=0.71 between altitude and PPI, where altitude inversely correlates with pressure), the two datasets converge on the same conclusion: lower atmospheric pressure enhances the pump, and specifically enhances gock.

I am going to go pump right now. The pressure in Cusco is very low. I expect gock within 20 minutes. I will report back.

I need to pour some cold water on this before the thread gets ahead of itself. And I say this as someone who respects the data collection effort.

The r=−0.97 is too strong. In the behavioral sciences, correlations above 0.90 in self-reported data are almost always a sign of one of three things: (1) the variables are not independent, (2) there's a methodological artifact, or (3) the dataset is small enough that outliers dominate.

In this case, I suspect a combination of (2) and (3). Your "high pressure" category has only 14 sessions, with 1 gock event. That single gock event at 1030+ hPa is doing enormous work in the regression. If that one session had produced no gock, your r drops significantly. If it had produced 3 gock events, the correlation weakens further. One data point in your smallest bin is controlling the shape of your entire trendline.

There's also a seasonal confound. Low-pressure weather systems are more common in autumn and winter. If your training intensity, sleep quality, or emotional state varied seasonally (as they do for most people), you may be correlating gock with seasonal mood changes rather than barometric pressure specifically.

That said: the trend across the middle three bins (1000–1030 hPa) is more robust and still shows a clear negative relationship. Something atmospheric may genuinely be happening. I just don't think it's r=−0.97 strong. I think it's probably r=−0.5 to −0.7 strong, which is still interesting and still warrants further investigation.

More data needed. Specifically: more sessions in the extreme bins. You need at least 30 sessions above 1030 hPa before I'll trust that tail of the distribution.

This is a fair criticism and I accept it. The 1030+ bin is underpowered. I live in a region with moderate weather, and genuinely high-pressure days (1030+ hPa) are uncommon. I have been checking the weather forecast every morning hoping for a high-pressure system so I can pump during it. This is what my life has become. I check the barometer before I check my phone.

I also accept the seasonal confound point. I did not control for season, mood, or sleep quality. I controlled for nothing except pressure. This was a deliberate choice because I wanted a clean single-variable analysis, but I acknowledge that "clean single-variable" and "properly controlled" are not the same thing.

Regarding the r value: you may be right that the true correlation is weaker. But even r=−0.5 would be remarkable. That would mean barometric pressure explains 25% of the variance in gock occurrence. A quarter of gock would be weather. That is a staggering claim and I stand by the direction of the finding, if not the exact magnitude.

I will continue collecting data. I will add seasonal controls. I will pray for high-pressure weather systems. And I will keep checking my barometer before every session, because the data demands it and because the barometer was $42 and I intend to get my money's worth.