ARTEMIA SALINA 335 



ARTEMIA SALINA. 



Report of the Committee appointed to investigate the progressive adaptation 

 to new conditions in Artemia salina (Prof. R. A. Fisher, F.R.S., 

 Chairman ; Dr. A. C. Faberge, Secretary ; Dr. F. Gross, Mr. A. G. 

 Lowndes, Dr. K. Mather, Dr. E. S. Russell, O.B.E., Prof. D, M. S. 

 Watson, F.R.S.). 



An outline of the programme of these experiments for the year 1937-38 

 was given in the last report of this Committee, presented at the Nottingham 

 session of the British Association. This programme has been followed in 

 all essentials. 



A total of 27,440 nauplii have been tested, representing an amount of 

 work considerably in excess of any one previous year. The testing and 

 breeding of the material has been carried out almost entirely by Miss S. B. 

 North. 



The distribution of the tested nauplii among the six generations and the 

 seven lines used this year is shown in Table L The data for lines Cg, C3, 

 C4, Ce, and C, are supplementary to those of the previous year. In 

 addition, two new lines have been started, the reciprocal crosses of Cg and 

 C7, two lines which had been selected through five generations last year : 

 Cg is a line which had shown a particularly strong and steady improvement 

 in resistance. 



It has been pointed out in the two previous reports that the chief factor 

 vitiating the precision of the results is an excessive discrepancy between 

 different broods of the same mating and generation. The first step taken 

 to overcome this dijfficulty was the system of testing nauplii in six diflferent 

 grades of poison solution rather than in only one. This device proved 

 inadequate to overcome the trouble, and in the present year untreated 

 controls were used. Each brood is divided into eight approximately 

 equal parts. Six of these batches are tested in six strengths of sodiimi 

 arsenite, as previously, and the other two batches are placed in medium 

 without poison, but are otherwise treated alike. By this means variations 

 from brood to brood in the natural death-rate during the testing period can 

 be taken into account ; though at the expense of much more complicated 

 calculations. 



It is clearly apparent from the results that the wide discrepancies between 

 parallel broods mentioned above is not due to differences in natural death 

 rate, but to variation in susceptibility to poison. An example of this, taken 

 from the third selected generation of line C3 is given in Table II (for the 

 meaning of the letters designating sodium arsenite solutions strengths see 

 Table III). 



Thus these controls are not sufficient to eliminate the excessive hetero- 

 geneity, and other means must be sought to improve the precision of the 

 experiment ; variation in temperature has been found to have no appreciable 

 effect. 



In order to utilise the data made available on natural death an entirely 

 new method of statistical reduction has had to be developed, which 

 involves fitting three parameters to the data. The first is the slope of the 

 probit regression line, i.e. the variability of susceptibility zvithin a brood. 

 This parameter is approximately constant for all broods of one line- 



