On the 3rd of April, 1999, a plan to pray for peace in Kosovo for two minutes every day for a month was announced. A ceremonial introduction was given on the internet, and some public service announcements augmented the word-of-moutn invitation to join with others praying at the same time. The time set was 19:00, universal time, which corresponds to 12:00 noon in California, where (I believe) the organizational source was centered. It was brought to my attention by Suzanne Taylor (of MightyCompanions), via an email note:

"This is to bring to your attention a 2-minute a day prayer for the Kosovo situation. It is taking place at noon PST, simultaneously all over the world. Marianne Williamson was the spokesperson who announced this, in an hour long broadcast from the Internet on Saturday. It is to go on for one month, which makes the end date May 3".

This constitutes a nicely defined, albeit implicit prediction for the GCP, from an independent source. It was not added to the formal prediction registry, but it nevertheless provides a very useful and informative example of a particular sort of prediction that has been made previously on a smaller scale. George deBeaumont processed the data according to the following pre-defined recipe:

1) Retrieved (web based extract form) data 4/3/99 thru 5/3/99 at 19:00:00 to 19:01:59 for each day.
2) Daily "online" eggs ranged from 10 to 15 over the period.
3) For each egg, summed the per second data for the full 2 minute period. This becomes the basic unit for all further calculations.
4) Calculated z^2 using theoretical std. dev. of 77.4597 (based on 24,000 trials per 2 minute interval and probability of 0.5).
5) Calculated cummulative z^2 - df.
6) Plotted cummulative z^2 - df verses df.

The following figure, By George deBeaumont, shows the result. The trend is non-significant, with a terminal probability of 0.193 at 360 df.