The verification of this ingenious hypothesis depends upon a variety of intricate meteorological conditions, some of which have been adversely interpreted by competent authorities.[902] What is still more serious, its acceptance seems precluded by time-relations of a simple kind. Dr. Wright[903] has established with some approach to certainty that glacial conditions ceased in Canada and the United States about ten or twelve thousand years ago. The erosive action of the Falls of Niagara qualifies them to serve as a clepsydra, or water-clock on a grand scale; and their chronological indications have been amply corroborated elsewhere and otherwise on the same continent. The astronomical Ice Age, however, should have been enormously more antique. No reconciliation of the facts with the theory appears possible. The first attempt at an experimental estimate of the "mean density" of the earth was Maskelyne's observation in 1774 of the deflection of a plumb-line through the attraction of Schehallien. The conclusion thence derived, that our globe weighs 4-1/2 times as much as an equal bulk of water,[904] was not very exact. It was considerably improved upon by Cavendish, who, in 1798, brought into use the "torsion-balance" constructed for the same purpose by John Michell. The resulting estimate of 5·48 was raised to 5·66 by Francis Baily's elaborate repetition of the process in 1838-42. From experiments on the subject made in 1872-73 by Cornu and Baille the slightly inferior value of 5·56 was derived; and it was further shown that the data collected by Baily, when corrected for a systematic error, gave practically the same result (5·55).[905] M. Wilsing's of 5·58, obtained at Potsdam in 1889,[906] nearly agreed with it; while Professor Poynting, by means of a common balance, arrived at a terrestrial mean density of 5·49.[907] Professor Boys next entered the field with an exquisite apparatus, in which a quartz fibre performed the functions of a torsion-rod; and the figure 5·53 determined by him, and exactly confirmed by Dr. Braun's research at Mariaschein, Bohemia, in 1896,[908] may be called the standard value of the required datum. Newton's guess at the average weight of the earth as five or six times that of water has thus been curiously verified.