A cursory glance at modern industry might lead one to the mistaken impression that energy consumption per capita has been rising steadily for centuries. This is far from the case. As usual, different statistical series tell a somewhat different story, but all point to a decrease in energy consumption per person through the internar period. Schurr (1960) has compiled statistics on energy use 1850–1955. Use of mineral fuels and hydroelectricity increases 180 times over the entire period. If, however, wood is included in the total, there is only a 17-fold increase, due to the much greater importance of wood as a fuel in the mid-nineteenth century (wood use naturally presents greater difficulties in estimation). Per capita increases over the period for the two series are 25 times or 2.5 times, respectively. Relative to GDP, the increase is 5 times or .5 times. If we include wood then, the ratio of energy use to output actually declined 50% over the century. Nor was this decline steady. The ratio had in fact increased to the 1910s and then fell. In per capita terms, energy use had risen from about 1885 to World War One, fell in the early 1920s, stagnated in the late 1920s, and fell again in the early 1930s, before rising back to the 1930 level in 1940. Decreased economic activity can provide a partial explanation for the early 1930s, but not for the 1920s. Energy consumption relative to GNP fell 15% in the 1930s (and a further 9% before 1955). Schurr points to four factors which drove the secular trends and especially internar experience. There was a structural shift away from heavy toward light industry. Productivity advance (both labor and capital) naturally decreased the energy requirement per dollar of output (and thus has a labor-saving effect not captured in the original industry). Technological advance caused great increases in the thermal efficiency of energy utilization (see below). Electrification encouraged an improvement in the ratio of energy input to output (Schurr, 1960:14–6, 145–56).