We cudgel our brains in vain. Perhaps the remainder of the sentence will help us. "The temperature is 30° Fahrenheit." "The unknown thing is thirty degrees." What then is a degree? That is the next question. When the Theory of Heat went out from sensation and left it behind, one of its first landing places was in the expansion of liquids--as in thermometer tubes. Here for some time was thought to be a satisfactory register of "temperature." But before long it became apparent that the degree--Fahrenheit, Réaumur, or what-not--was an entirely arbitrary thing, also that it was not the _same_[28] thing at one end of the scale as the other, and finally that the scale itself had no starting point! This was awkward, so a move was made to the air thermometer, and there was some talk about an absolute zero and absolute temperatures; it was thought that the Unknown thing showed itself most clearly and simply in the expansion of air and other gases, and that the "degree" might fairly be measured in terms of this expansion. But in a little time this kind of thermometer--chiefly because no gas turned out to be "theoretically perfect"--broke down, absolute zero and all, and another step had to be made--namely, to the dynamical theory. It was announced that the Unknown thing might be measured in terms of mechanical energy, and Joule at Manchester proclaimed that the work done by any quantity of water falling there a distance of 772 feet is capable of raising that water one degree Fahrenheit.[29] Here seemed something definite. To measure temperature by mass and velocity, to measure a degree by the flight of a stone, or the heat in the human body by the fall of a factory chimney--if rather roundabout and elusive of the main question--seemed at any rate promising of exact results! Unfortunately the difficulty was to pass from the theory to its application. The complicated nature of the problem, the "imperfection" of the gases and other bodies under consideration, the latent and specific heats to be allowed for, the elusive nature of heat in experiment, and the variable value of the degree itself--all render the conclusions on this subject most precarious; and the general equations connecting the Fahrenheit or other temperatures with a thermo-dynamic scale--while they become so unwieldy as to be practically useless--are themselves after all only approximate.