Nine USGS deep earthquakes in 2016-2025 did not resolve an April-September excess

We tested whether the April-September excess reported in 2015 for large deep earthquakes persisted in 2016-2025. We used global USGS events with preferred magnitude at least 7.0 and hypocenter depth greater than 500 km. Five of nine events occurred in April-September. The fraction was 0.556, with an exact 95% interval of 0.212 to 0.863 under independent sampling. The one-sided test against uniform calendar time gave p = 0.502. A test that kept event phases together within each year gave p = 0.502. The result does not resolve an excess or exclude the earlier reported fraction of 0.70. Power at that fraction was only 19.6%. An independent-event sample of 40 is the smallest with at least 80% exact-test power. GCMT gave five of seven events before magnitude rounding and five of nine after rounding. Two November 2022 events explain the difference. Small counts, event dependence, current catalog revisions and threshold definitions limit the test. The package includes event records, a fixed test plan, code, source hashes, all checks and a figure. It establishes no physical cause and does not falsify the seasonal prediction.

deep earthquakes · seasonality · prospective test · catalog sensitivity · statistical power