RECENT experiments by Fenn(l, 2) have shown that when a muscle is allowed to shorten and do work there is a further output of energy beyond that liberated in an isometric contraction. When the load was held by the muscle during contraction only, and not during relaxation, the total extra energy liberated was approximately equal to the work done, so that the energy liberated as heat was constant. If the load rested on the muscle during relaxation then still more heat was given out. It seemed desirable to consider the matter in greater detail, and to make an attempt to follow the course of the heat production during the contraction by analysing the curve of galvanometer deflection usinag the method explained elsewhere(3). By such means it was hoped to determine in which phase the extra energy associated with work was liberated. With a view to this, and to other measurements of the kind, considerable improvements have been made in the technique.The thermopile, constructed by soldering, has been improved by making it in one sheet, zig-zag fashion, holding each outer junction by thread to the supports, instead of winding the wire either round a sheet of insulator or round the arms of a U-shaped frame. If the same wire were used the new method would give only half the number of junctions per unit length, but by using finer wire (36 S.W.G. iron and constantan) the number of junctions per unit length has been maintained (35 to 40 junctions per centimetre) without making the electrical resistance excessive. Using as little insulation as possible the thermal capacity of the part of the thermopile under the muscle is now much less than before, so that its temperature changes follow more quickly those of the muscle.A further great improvement has been made by introducing a new suspension into the Paschen galvanometer used. This magnet system was constructed, with much stronger magnets of cobalt steel, by