Coon’s Agar and Prune Agar one would obtain a total of 723 colonies, whereas the same suspension on Cook’s Agar and Czapek’s Agar would give only 461, and the calculated numbers of fungi per gram of soil would be totally different. Further, if a single medium be taken, it is found that slight alterations in the degree of acidity may make very considerable differences in the final numbers. Thus Coon’s Agar acidified to a hydrogen ion concentration of 5·0 gave as the results of four series the following average numbers of colonies per plate, 17, 23·75, 18, 23. When, however, the medium was acidified to a PH of 4·0 to 4·3, corresponding averages from three series were 38, 46·3, and 44·8; i.e. the final estimations of numbers of fungi in the soil was about twice as great. Again, the degree of dilution of soil suspension used in plating may also be a very serious factor. Thus, if a series of dilutions be made of 1/80,000, 1/40,000, 1/20,000, 1/10,000, 1/5,000 and 1/2,500, the average plate numbers should be in the proportions of 1, 2, 4, 8, 16, and 32 respectively. In an actual experiment, the following average plate numbers were obtained, 15·4, 32·8, 59·1, 104·0, 150, 224·5, which show a very decided reduction in the higher numbers. If, however, dilutions of a suspension of spores of a single species be made, this reduction does not occur. These are but three of the very numerous factors involved in the technique of quantitative estimation, and every single factor may be the source of errors of similar magnitude, minute fluctuations in the operations leading to the final platings having very considerable effect upon the numbers of colonies that develop. By critically evaluating each particular factor in the method, and making statistical correction, it has, however, been found possible to obtain series of duplicate plates comparing very favourably and thus to extract certain figures which, whilst not possessing any final value, have yet a certain general and comparative worth. Thus, 20·0, 18·2, and 16·8 were obtained as the averages of six plates each, of a soil suspension divided into three parts, and the individual plate numbers in all three series were within the range of normal distribution. The meaning of these numerical estimates in relation to fungi per gram of soil sample is, however, entirely hypothetical, and to have value quantitative comparison should only be made between single species or groups of species closely related physiologically, and where the technique is standardised. [Illustration: FIG. 19.--Monthly Counts of Numbers of Fungi per gramme of Dry Soil. Broadbalk Plot 2 (Farmyard Manure), Rothamsted. X-axis: ~Apr.~ 1921 May Jun. ~Jul.~ Aug. Sep. ~Oct.~ Nov. ~Dec.~ Jan. 1922 ~Feb.~ Mar. Apr. May ~Jun.~ Jul. Aug. ~Sep.~ ~Oct.~ Y-axis: 10.000 per Gramme of Soil]