In recent (historically dated) lava flows, lava which hardened in air may date to zero. It may also date up to 1 million years old. (1) The reason for this discrepancy is not clear. It has been suggested that some of the age may be correlated with the original makeup of the argon in the basalt before it erupted. Parts which cooled under water may date much older, even over 40 million years (2), because of "premature" solidification (quenching; hardening before all the argon came out). So different parts of the same recent lava flow have different dates.

Dalrymple's dates are biased by (1) using 296.1 instead of 295.5 as the ratio of 40Ar to 36Ar, and (2) testing multiple times whenever a result was not statistically compatible with zero age, but testing only once whenever it was. The first bias was justified by Dalrymple because 296.1 was the 40Ar/36Ar ratio in the air at his laboratory in Hawaii, presumably because of volcanic activity in the area. While this might be reasonable, it should be applied to all samples analyzed at the laboratory, not just recent ones, to be valid. And with adequate removal of argon from the air, from samples, it would not be justified even then for samples from other localities. Changing the 40Ar/36Ar ratio back to 295.5 would appear to make at least one other result statistically different from air.

The second bias cannot be justified. It can only be excused as a mistake. If one is trying to accurately ascertain the proportion of dates that are different from zero, one should test all samples using the same protocol. The retesting of all samples that are statistically different from zero, while not retesting those that are not, biases the results in favor of those that are not statistically different from zero. In this connection, it is interesting to note that all the retested samples were still statistically different from zero, although individual dates from those samples were sometimes indistinguishable from zero. Only one sample which was different from zero was not retested, and the article states that this was only because the laboratory ran out of sample.

References:

1. G. B. Dalrymple: "40Ar/36Ar analysis of historic lava flows." Earth Planet Sci Lett 1969;6:47-55. See also A. A. Snelling ("The Cause of Anomalous Potassium-Argon "Ages" for Recent Andesite Flows at Mt Ngauruhoe, New Zealand, and the Implications for Potassium-Argon "Dating"." R.E. Walsh, ed., Proceedings of the Fourth International Conference on Creationism. 1998, Pittsburgh, PA, Creation Science Fellowship, pp. 503-525) for a further collection of dates.

2. G. B. Dalrymple and J. G. Moore: "Argon 40: Excess in submarine pillow basalts from Kilauea Volcano, Hawaii." Science 1968;161:1132-5. See also C. S. Noble and J. J. Naughton: "Deep-ocean basalts: Inert gas content and uncertainties in age dating." Science 1968;162:265-7.

______________________________________________________