The Distribution of Leucocytes on the Counting Chamber.

It is almost a statistical truism that the number of cells on each square of a haemocytometer is distributed in a Poisson series, but in fact it is easy to show that the distribution departs considerably from the theoretical expectation. The reason, perhaps, is that the experimental work confirming this theory was done on the old well type of haemocytometer, whereas the Burker counting chamber is now universally used, and adds new and less predictable variations to random sampling. My attention was drawn to this point by a statistically controlled series of leucocyte counts on Biirker counting chambers with the Neubauer ruling. I counted, as is usual, the number of cells on four areas of one square millimetre at the corners of a square with a side of three millimetres, and recorded the four sub-totals separately. Over the whole series, the two squares nearest to the.filling point of the chamber gave nearly identical totals 3.5 per cent below the mean value for all four squares, and the two further squares similarly gave nearly identical totals 3.5 per cent above the mean value. The difference was statistically highly significant. It seemed worth while to test the extent of this deviation from the theoretical Poisson distribution, and to see whether it might be a source of error in blood counts.