time if the above precautions are followed. =47. Location of readings.= Samples should invariably be taken in spots which are both typical and normal, especially when they are to be used as representative of a particular area or habitat. A slight change in slope, soil composition, in the amount of dead or living cover, etc., will produce considerable change in the amount of water present. Where habitat and formation are uniform, fewer precautions are necessary. This is a rare circumstance, and as a rule determinations must be made wherever appreciable differences are in evidence. The problem is simpler when readings are taken with reference to the structure or modifications of a particular species, but even here, check readings in several places are of great value. The variation of water in a spot apparently uniform has been found to be slight in the prairies and the mountains. In taking three samples in spots a few inches to several feet apart, the difference in the amount of water has rarely exceeded 1 per cent, which is practically negligible. Gardner[2] found that 16 samples taken to a depth of 3 inches, in as many different portions of a carefully prepared, denuded soil plot, showed a variation of 7½ per cent. This is partially explained by the shallowness of the samples, but even then the results of the two investigations are in serious conflict and indicate that the question needs especial study. It should be further pointed out that all readings should be made well within a particular area, and not near its edge, and that, in the case of large diversified habitats, it is the consocies and the society which indicate the obvious variations in the structure of the habitat.