But when we are watching the splash made by the fall of a liquid drop, instead of a solid sphere, there is a new and more serious source of difficulty. For the drop as it lies on the smoked glass cup is not perfectly spherical, but is flattened by its own weight, as shown in Fig. 3, and on the sudden removal of the supporting cup it oscillates between an oval form, elongated vertically, and a flattened form (see Fig. 4). These oscillations are unavoidable, and their extent will depend partly on the amount of adhesion between the smoked surface and the drop, and as this adhesion is never entirely absent and is variable, depending partly on the length of time that the drop has been lying in the cup, it follows that the drop will always receive a slight tug downwards at starting, which will be greater on some occasions than on others. On this account not only will the time taken to reach the water vary slightly, but the drop will strike it sometimes when elongated and sometimes when flattened, and the resulting splash will be affected by this circumstance. [Illustration: FIG. 3] [Illustration: FIG. 4] The four photographs on the next page were taken in succession in order to afford the reader an opportunity of judging for himself the sort of accuracy attainable when a liquid drop was concerned. The fall was 30 centim., and the greatest discrepancy is 4·8 millimetres, corresponding to 1/560 of a second. Thus even here the error does not amount to two-thousandths of a second. [Illustration: Photographs taken to test the timing of a falling drop. 1 2 3 4] With higher falls the timing sphere is moving more quickly past the discharging knobs, and the error due to a longer or shorter spark is correspondingly less, so that it appears safe to say that the accuracy of the timing was such that, when all precautions were taken, any desired stage could be picked out within two-thousandths of a second. It is not however pretended that the precautions necessary for the most accurate timing were always taken, especially in the earlier Series of Photographs, for the main object of the experiments was to find out what happened, and only incidentally to ascertain exactly how long it took to happen, and there is no doubt that on some occasions, through the smoke-film being allowed to wear away, adhesions to the dropping cup occurred, with a corresponding disturbance of the timing, before the defect was noticed and remedied. [Illustration: FIG. 5 Photograph of the edge of a rapidly whirling disc.]