About 1876 the author had occasion to devise a method of indicating the rate of flow of a liquid through a pipe system, without interfering with the flow. For this purpose he used a ball, heavier than the liquid, in a slightly tapered glass tube, through which the liquid flowed from the narrower to the wider end. In the first instance the tube was set vertical, with the wider part above. The ball fitted more or less closely at the lower end: it was carried up by the current till the clearance round it became such as to suit the particular rate of flow. Its height was read off on a scale fixed alongside of the tube, and served to indicate the rate of the flow, once the relation of scale-reading to flow had been experimentally determined.