By T.S.H. EYTINGE, Cainsville, Canada.

It is well known that the sun's distance has been determined from the velocity of light. It has been found, by terrrestrial experiments, about how fast light travels, and, knowing from certain astronomical phenomena the time light requires to pass from the sun to the earth, we have been able to determine the sun's distance.

There are several methods of determining the velocity of light, but hitherto only two plans have been used to detect the time light occupies in passing from the sun to the earth. This time was first discovered by observations of the satellites of Jupiter. It was found that the interval between the eclipses of these bodies was not always the same--that the eclipses occurred earlier when Jupiter was nearest the earth, and later when he was at his greatest distance. Roemer, a Danish astronomer, first detected the cause of this variation. The second method by which this time has been found is the aberration of stellar light. This refined method was detected by the great English astronomer Bradley.

About two years ago it occurred to me that a third method can be used to solve this important problem. My plan is this: It is well known that many variable stars, such as Algol, σ Librae, U Coronae, and the remarkable variable D.M. + 1.3408°, discovered by Mr. E.F. Sawyer, fluctuate at regular intervals. Now, I believe it is possible to determine very accurately the intervals between these changes, and, by noting the change of time in these intervals, when the earth is in different points of its orbit, we get the time light requires to cross that orbit. For, as in the case of the satellites of Jupiter, when the star is "in opposition," the changes will occur earlier than when it is in conjunction or approaching that point. I have recently put this plan to the test, and hope before long to make known the results.

In detecting the changes of variables, I have attempted to substitute, in place of the ordinary eye observations, a very delicate thermopile, which registers the changes in the star's heat. So far as I know, this is the first application of the thermopile to variables.