312 Proceedings of Royal Society of Edinburgh. [june 20, 
sists in mixing a sample of water of definite bulk with a quantity of 
liquefied nutrient jelly previously sterilised. This mixture is poured 
on a sterile glass plate with aseptic precautions. When it solidifies, 
the microbes in it are fixed and develop, each becoming the centre of 
a colony growing in the jelly, and each colony representing a “ pure 
cultivation ” of its parent germ. The number of such colonies 
bears a definite relation to the total number of microbes present in 
the original sample of water, and their varied appearances show with 
what diversity of species the sample was inhabited. Species which 
cannot be recognised in this way at once may be identified by re- 
moving a portion of one such colony to a separate quantity of 
culture material, in which its characteristic growth may be separately 
observed. A convenient form of apparatus is shown in Plate X. 
It consists of a deep glass bell of 8 inches inside diameter, standing 
in a glass dish that closely fits its mouth. Within are alternate 
square glass plates and india-rubber washers, fitting closely to the 
inside of the bell-jar. When the apparatus is closed it will travel 
safely without any movement of the plate, and can thus be conveyed 
to the near vicinity of the water which it is desired to test. The 
plates are sterilised by heating at 170° C for one hour, and 
the rings by steeping them in 1 per cent, aqueous solution of 
per chloride of mercury for twenty-four hours. A piece of thick 
filter paper soaked in the same solution is placed in the floor of the 
apparatus. A series of ten or twelve plates is contained in the 
apparatus. In making observations the upper two or three of these 
should be used as “ control ” plates, since they run the greatest risk of 
aerial contamination from their longer exposure to the air in manipu- 
lating the apparatus. They are charged with a layer of the 
nutrient jelly alone without the addition of the water to be tested. 
If they give negative results, i.e. if the layer of nutrient jelly 
upon them shows no foci of microbic growth, the manipulative 
procedure in the experiment may be considered reliable. In 
examining a river for microbes at different parts of its course, 
one such set of twelve plates may advantageously be used at each 
point, a measured quantity of the water being used in each case, 
so that the experiments may have uniformity of scale throughout. 
Two such experimental plates are shown in Plate XI., in which 
the results of the test are shown in the case of a river examined 
