At the points of support in a line of columns, the beams should be free from all compressive stress, transmitted through the lines of columns from floors above, by means of iron pintles between the cap of one column and the floor of the next one carrying this load. A faulty method of construction, quite frequently used, consists in covering each column with a bolster of timber, four or five feet long, reaching out under the floor beams. The transverse contraction of wood in seasoning after it is in position in the mill varies from three-eighths of an inch to double that quantity per foot; and the aggregation of such shrinkage amounts to a very considerable distortion or settling of the floor in a mill of several stories. Moreover, the resistance of timber to transverse crushing has been shown by experiments on the testing machine at the United States arsenal at Watertown to be about three times the resistance to longitudinal crushing. Iron columns for mills have been entirely displaced by those of timber, as it was found that the latter were more reliable in resistance to fire, were freer from defects in construction, and possessed less tendency to vibration. A series of tests on full-sized mill columns of various forms of construction and age, made in the experiments referred to, at the Watertown arsenal, showed that resistance to crushing of Southern pine columns was about 4,500 pounds to the square inch, and remarkably uniform as to the different results. In white oak there was a wider range, owing to the difference in the grain of the various samples, the generality of the specimens being of somewhat less resistance than that of Southern pine. It was furthermore found by these experiments, on comparing the crushing resistance of a full-sized column with that of a portion of the same, perhaps two feet in length, that the results were practically identical, likewise that within the limits of construction used for these columns the question of flexure did not enter at all in the problem, but they gave way by direct crushing, and that the resistance to crushing was proportional to its load upon the minimum cross section. The precedents of safe construction in this matter show that wood columns in mills have successfully sustained for many years a load of six hundred pounds to the square inch without deterioration. As the resistance of such columns is proportional to the cross section, the results of these experiments have changed the practice of mill engineers in the matter; and square columns are of almost universal use, which interfere with no greater area on the floor than the round column of the same diameter, while they furnish an increased resistance of a little over twenty per cent. in excess.