if moving with the velocity of light would have kinetic energy equal to 6.7 x 1016 ergs. Hence we conclude that in each gramme of radium at least ⅛ of a milligramme, i.e., about 1 part in 8,000, must be in the ether. Considerations of this nature induced me some time ago to make experiments on radium to see if I could get any evidence of part of its mass being of an abnormal kind. The best test I could think of was to see if the proportion betwen mass and weight was the same for radium as for ordinary substances. If the part of the mass of radium which is in the ether were without weight then a gramme of radium would weigh less than a gramme of a substance which had not so large a proportion of its mass in the ether. Now the proportion between mass and weight can be got very accurately by measuring the time of swing of a pendulum; so I constructed a pendulum whose bob was made of radium, set it swinging in a vacuum and determined its time of vibration, to see if this were the same as that of a pendulum of the same length whose bob is made of brass or iron. Unfortunately radium cannot be obtained in large quantities, so that the radium pendulum was very light, and did not therefore go on swinging as long as a heavier pendulum would have done; this made very accurate determinations of the time of swing impossible, but I was able to show that to about 1 part in 3,000 the time of swing of a radium pendulum was the same as that of a pendulum of the same size and shape made of brass or iron. The minimum difference we should expect from theory is 1 part in 8,000, so that this experiment shows that if there is any abnormality in the ratio of the mass to weight for radium it does not much exceed that calculated from the amount of heat given out by the radium during its transformation. With larger pendulums the value of the ratio of mass to weight can be determined with far greater accuracy than 1 part in 8,000; for example, Bessel three-quarters of a century ago showed that this ratio was the same for ivory as for brass to an
