GERMAN PLANKTON INVESTIGATIONS. 301 



plate is used, and this result, combined with the previous 

 one, gives us a fresh average. 



When this second average does not differ from the first 

 by more than 5 per cent., the result may be taken to be 

 satisfactory. In the form appended we find for Melosira 

 the numbers 43, 38 and 48 on the first three plates. This 

 gives us as average 43. The next counting gives us 38. 

 The total now becomes 167 and the average 42. 



Then it follows that 43 : 100 = 42 : x. 



x = 97-7. 

 That is, the difference is 2*3 per cent., and therefore the 

 counting can be regarded as accurate enough and com- 

 pleted, but of course it is always more advisable to use 

 too many plates than too few. Having thus arrived at a 

 sufficiently accurate result, we do not estimate the 

 Diatoms on succeeding plates, and we can therefore use a 

 weaker dilution and magnification. For the rarer forms 

 the first dilution can be used, and from this 1 ccm. and 

 finally 2'5 ccm., counted through ; this work is much 

 quicker, as a smaller magnification is used and only a 

 few animals have to be counted. Each separate counting 

 is noted on a form, an example of which is given (pp. 302-3). 

 The forms originally used by Hensen are very compli- 

 cated, and I have not attempted to produce one in its 

 entirety ; still the appended form is sufficient for all 

 practical purposes. It may here be mentioned that the 

 criticisms of Kofoid and the results obtained by Lohmann 

 (9) seem to render some modification of the original form 

 necessary. It seems useless to include organisms in the 

 calculation the majority of which are proved to slip 

 through the silk bolting cloth, No. 20. The forms are of 

 two kinds. One is used for entering the results of an 

 individual catch ; the other for comparing such catches 

 together. An example of the first only is appended. This 



