Some statistical “distributions” which, when plotted on an “arithmetical” basis, do not in the least resemble a normal or Gaussian distribution, become very similar to one when plotted on a logarithmic or “geometrical” basis. In this paper I examine in particular the difference of two declining exponential curves, and, as a special case, a simple single declining exponential. On a logarithmic base these curves are somewhat skewed, but none the less are very close to a lognormal, and nearly all the variance can be accounted for (something like 99%), but there seems to be a limited range of “standard deviations” that can be accommodated. In human ecology, for instance, women's age at first marriage, which among western nations is known to be close to a lognormal, can be slightly better fitted by the difference of two declining exponentials, and a physical interpretation can be given to the latter, though not so easily to the former. The same thing is true of the failure of industrial products under repeated impact, or under prolonged stress. The apparently lognormal distribution of commonness and rarity differs from these in having a much greater value of the standard deviation, and it is not so easily explained. This may be because it is, in fact, a totally different phenomenon, but there is also a possibility that some modification or extension of the present paper may point the way to an explanation.