We propose a modification to a commonly used technique for lifetime determination of solar features. Suggestions for practical implementation of the method are presented. In addition, a serious source of bias in lifetime determination from time-ordered data is discussed.
A commonly used technique for lifetime determination of solar features consists of tagging the features on a photograph contained in a time-ordered set of photographs and tracing the features back to birth and forward to disappearance (see, e.g., Janssens (1970) and references therein). This produces a curve of number surviving as a function of time with a peak in the middle at t = 0. The quoted lifetime value is then usually taken to be the full width at half maximum of this curve, or the mean lifetime of all features seen on the photo at t = 0.
If the features in question (e.g., supergranule cells) actually belong to a, single class and can each be considered independent events, then the FWHM overestimates the lifetime by a factor of 2 In 2, or about 39%. What has actually been measured is twice the 'half period' T, which is related to the lifetime by (see, e.g., Evans, 1955, pp. 470 if):
In 2 T= h
The quantity /~--1 is the lifetime and the measured data will behave according to the radioactive decay law (Bateman, 1910) N(t) = Noe -~t.