Now, previous writers have come to a different conclusion from the same general line of argument, because they have started with different assumptions. Professor Newcomb, whose statement made some years back is usually followed, assumed 100 million stars each five times as large as our sun, equal to 500 million suns in all, and he distributed them equally throughout a sphere 30,000 light-years in diameter. Thus he has half the amount of matter assumed by Lord Kelvin, but nearly five times the extent, the result being that gravity could only produce a maximum speed of twenty-five miles a second; whereas on Lord Kelvin's assumption a maximum speed of seventy miles a second would be produced, or even more. By this latter calculation we find no insuperable difficulty in the speed of any of the stars being beyond the power of gravitation to produce, because the rates here given are the direct results of gravitation acting on bodies almost uniformly distributed through space. Irregular distribution, such as we see everywhere in the universe, might lead to both greater and less velocities; and if we further take account of collisions and near approaches of large masses resulting in explosive disruptions, we might have almost any amount of motion as the result, but as this motion would be produced by gravitation within the system, it could equally well be controlled by gravitation. [Illustration: DIAGRAM OF STELLAR UNIVERSE (Plan). 1. Central part of Solar Cluster. 2. Sun's Orbit (Black spot). 3. Outer limit of Solar Cluster. 4. Milky Way.] In order that my readers may better understand the calculations of Lord Kelvin, and also the general conclusions of astronomers as to the form and dimensions of the stellar universe, I have drawn two diagrams, one showing a plan on the central plane of the Milky Way, the other a section through its poles. Both are on the same scale, and they show the total diameter across the Milky Way as being 3600 light-years, or about half that postulated by Lord Kelvin for his hypothetical universe. I do this because the dimensions given by him are those which are sufficient to lead to motions near the centre such as the stars now possess in a minimum period of twenty-five million years after the initial arrangement he supposes, at which later epoch which we are now supposed to have reached, the whole system would of course be greatly reduced in extent by aggregations towards and near the centre. These dimensions also seem to accord sufficiently with the actual distances of stars as yet measured. The smallest parallax which has been determined with any certainty, according to Professor Newcomb's list, is that of Gamma Cassiopeiæ, which is one-hundredth of a second (0".01), while Lord Kelvin gives none smaller than 0".02, and these will all be included within the solar cluster as I have shown it. [Illustration: DIAGRAM OF STELLAR UNIVERSE (Section). Section through Poles of Milky Way.]