It is marvellous that under such circumstances they should have continued the work, since it was evident, from the pieces twice supplied, that the progress of the settlement was going on sufficiently to make itself sensible, even while the centering of the dome remained; yet it does not appear, as I have already said, that any immediate mischief followed the striking of those centres, and it was not till 1796 that the ultimate stability of the edifice was considered doubtful. At that time a commission of architects was appointed to examine the state of the building, and report on the best means of proceeding. These gentlemen examined the piers, and completely ascertained the defective mode of workmanship which I have above explained; and they found that the piers and columns under the dome, had settled irregularly in consequence of it. One pier had sunk five inches and two lines, French measure, the whole of which must have taken place in the height of the columns (thirty-seven feet eight inches), as every thing above and below was firm. Such defects in the workmanship seemed sufficient to account for the failure of the construction; but it was necessary to know, whether if perfect, the piers would have had sufficient solidity, and whether there was any defect necessary to be attended to in the disposition of the weight above. Soufflot made some experiments to ascertain the pressure which the stone ‘_du fond de Bagneux_’ used in these pillars, would support; but it appeared probable that the instrument he used was defective. Rondelet therefore repeated the experiments, both with Soufflot’s machine, and with one of his own contrivance. According to the first, each pier would support a weight of seventy million three hundred and sixty-two thousand, seven hundred and twenty pounds, supposing it to be a single block of stone; according to the last, of twenty-seven millions, three hundred and twenty-nine thousand two hundred and twenty-two; a tremendous difference, and yet the estimate is still probably too high, as even in Rondelet’s machine, some power is lost by friction. As however it is probable, that from the bad construction of the piers, the weight was not supported by more than a fourth part of the superficies; their strength, calculated on Rondelet’s machine, would not exceed six million eight hundred and thirty-two thousand, three hundred and five pounds, while the weight of a quarter of the dome was ascertained to be seven millions, four hundred and forty-nine thousand, nine hundred and eighty. We must, however, be careful how we make use of these combinations of experiment and calculation, since it would appear from them that the piers of the bridge of Neuilly, to support arches of 120 feet span, instead of 13 feet thick, as they actually are, need only have been about four inches, and the walls of a house five stories high, require only three lines and a half in thickness at the bottom. As for the distribution of weight the commissioners condemned the method adopted, owing to a change in the plan during the progress of the work, of making the drum of the dome pass a little on the _outside_ of the line of the uprights; but they contented themselves with recommending the establishment of centering to relieve the weight, while the broken stones were removed, and replaced with such an incrustation carefully worked, as would be sufficient to sustain the whole building.