222 MUTATIONS 



stated, but one can assume with some reason that the amount probably 

 injected was distributed into body water. Assuming that this initial 

 concentration was maintained for a whole day, we can again obtain an 

 underestimate of the caffeine effect, which, however, should not be as 

 much in error as that from the experiment with larvae. 



The Drosophila results, unlike those obtained with the bacterial 

 chemostat, are not mutation rates, but final mutant frequencies in 

 sperm. To use our equation to calculate p, the probability of induced 

 mutation per cell per division cycle, we must have information about 

 mi, m2, etc., for Drosophila. From the number of germ cells at pupa- 

 tion and the size of the sperm clones (55) we can estimate that there 

 are about 30 cell generations in the lineage of the sperm. Since the 

 effect measured may presumably involve any of some 300 genie loci on 

 the X chromosome, we must let n = 300 in our equation. Finally, we 

 have to reduce the caffeine concentration to 1 mg/1. The experiment 

 with larvae then gives p = 5 X 10"^*^ as compared with 10"^ for the 

 T5 gene in E. coli. As we have pointed out, this figure is necessarily an 

 underestimate. 



For the single-injection experiment we assume a level of 700 mg/1 in 

 the body water of the fly. Here there was no exposure to caffeine during 

 the larval period; we will estimate that, on the average during this 

 experiment, there were 10 division cycles of stem cells and spermato- 

 cytes. Calculation then yields approximately p = 5 X 10"^. 



I think it is interesting that, rough as these estimates are, the 

 Drosophila data, based on a large number of different genes, lead to a 

 mutagenicity of caffeine that is not too very far from what the bac- 

 terial data say. No such experiments with caffeine have been done in 

 mice or any other mammalian organism. I should think that the only 

 way to get some sound evidence to replace these speculations would be 

 to initiate some good experiments in mice, to see whether the extrapola- 

 tions presented here have any validity at all.* 



* Addendum: Lyon et al. compared the specific locus mutation rates in male 

 and female mice which had 1 g/1 caffeine dissolved in their drinking water up 

 to the age of ten weeks, parents having been similarly treated from the time of 

 mating. The mutation rate per locus per gamete was 0.44 X 10"^ in males and 

 0.0 in females; these do not differ significantly from each other, nor from the 

 known spontaneous rates. In this experiment 64,000 progeny were examined. 

 However, radiation equivalent less than about 25 r would probably not have 

 been detected. Cattanach has obtained essentially negative results in studying 

 the possibility that caffeine at 2 g/1 in drinking water causes translocations in 

 mice. Lyon, M. F., R. J. S. Phillips, and A. G. Searle. A test for mutagenicity 

 of caffeine in mice. Z. Vererbungslehre, in press (1962). 



