524 On the Distrihution of the Means of Samples 
The actual number of bacteria in each sample of 25, 50 or 100 cells was copied 
out on a separate slip of cardboard, so that we had 800, 400 and 200 cards 
corresponding to the samples of 25, 50 and 100 cells. The cards used were small 
pieces of correspondence cards which, being smooth, could be shuffled very easily. 
With these cards the following experiments were performed. Taking the case 
of the 400 cards relating to the fifties, the cards were thoroughly shuffled and one 
drawn out. The number on the card was written down, the card not being 
replaced. The remainder were once more shuffled, another drawing was made and 
the process was repeated until the whole 400 had been drawn. The cards were 
then once more shuffled and the process repeated, the second set of numbers being 
written down one by one under the first set of values. Each number of the first 
set was then divided by the corresponding number in the second set and the 400 
Opsonic Index. 
Gbaph 8. Curve of lOO's secondary without replacement. 
