The following paper is one member of a series designed to elucidate the circumstances of correct time-keeping, the first two of which were published in 1918.

The problem has in the meantime gained appreciably in interest. It was known that time-determinations from night to night at the same observatory were liable to discrepancies, which were presumed to have an accidental character and would be eliminated by grouping. Now, comparisons by the aid of wireless telegraphy of determinations made at different observatories, and grouped with the intention of eliminating accidental error, exhibit like discrepancies, of a more systematic character, with a range certainly not less than 0s·3. The search for the cause of this fault is engaging the attention of astronomers. It is a larger fault than is found in any other region of their science; and as its source is obscure, additional value attaches to any collateral investigation bearing upon it.