Hi all;

Recently Jim Berkland posted a list of 215 great historical quakes going back to 1972 and claimed that a large percentage of them fell within his syzygy windows. That's an irresistible challenge so I looked into it.

I first looked at only the standard -1 to +6 day windows and the results were marginally significant. But Jim protested that I was ignoring perigee, which can shift the start of a window up to -3 day. Then he pointed out that Kurile quakes were known to not fit.

Making those changes raised his results to much higher levels but I found differences between his calculations and mine, leading me to wonder if there were transcription errors in the quake list.

Rather than check all 215 quakes against the half dozen sources he had used, I decided to do an independent study. I searched the ANSS and NEIC historical databases for the same time period. I eliminated the Kurile quakes and tested the remaining 197 quakes against his standard windows including the perigee adjustment.

Imagine my surprise when the z-binomial test gave him a 99.7% significance rating. That's based on 122 hits out of 197 quakes when the probability of success is 0.542 (based on window days vs total days).

Petra had sent me 216 of her predictions to test and the initial pass gave her a significance level which was literally off the scale. I consulted Drs Jones and Vidale, who suggested thet I should eliminate anything below 4.5 since the database is incomplete down there, eliminate aftershocks which have a higher probability and eliminate overlapping predictions which would give 2 hits for 1 quake. I don't usually bother with those tests because predictors normally don't work that way. But Petra had done all 3. Eliminating those elements lowered her significance drastically.

She was NOT pleased....

Then I tested the 137 predictions she has posted on her website and found they were below chance without eliminating anything.

Case closed for now.

Roger