There's been some speculation (e.g. at Arms Control Wonk) that the detection of short-lived fission products and activation products are evidence of intermittent nuclear criticalities. While it's been suggested that some isotope identifications are probably errors (computer confusion of one gamma-ray peak with another of very similar energy), one problem not resolved in the media is, how to explain the improbably high activities observed? For example, the erroneous measurement of 78 Curies/liter of iodine-134 in turbine hall puddles -- a fission product with a 53 minute half life. Yomiuri Shimbun reports TEPCO as concluding it was actually cobalt-56, or maybe cesium-134 (they backtracked twice). But such high activity levels of these longer-lived fission products would be even more confusing.
I suggest a simple explanation for this absurdity. Take a look again at this set of faulty measurements (leftmost column; I've appended this table at the end of this post). We see 2.9×109 Bq/cm3 (or 78 Ci/L) 134I, t1/2 = 53 minutes. But look at the time! (Top rows, first page). This sample was obtained at 8:50, and measured (by gamma spectrometry I speculate) at 18:50 of the same day -- 10 hours later. Eleven 134I half lives later.
So what's the meaning of the reported figure? To report the measured activity of 134I, after 99.96% of it had decayed away, would be mostly meaningless. I speculate that TEPCO had, rather, inferred the original 134I activity by taking the measured activity and working backwards. That is, compensated for 11.32 half-lives of decay by just multiplying them back in (a factor of 2,558x). This would be fine -- if the isotope really was 134I.
But suppose the true contaminant was really a long-lived isotope, half-life of months (Co-56) or years (Cs-134). Then the decay would have been negligible; and the back-calculation would have introduced an error factor of 2,558x. The actual activity would then have been 3 orders of magnitude lower than the reported one, or about 1.13×106 Bq/cm3 (31 mCi/L) of a long-lived isotope. Which would be consistent with the other measurements.
Update: Cyril R points out that tellurium-134, t1/2 = 41.8 minutes, β-decays to 134I, which would alter the calculations (by a small factor).