Before deciding on the best way to light the military workhouse at Munich, Rumford made a series of experiments on the relative economy of different methods, and for this purpose designed his well-known shadow-photometer. In the final form of this instrument the shadows were thrown on a plate of ground glass covered with paper, forming the back of a small box, from which all extraneous light was excluded. Two rods were placed in front of this screen, and the lights to be compared were so situated that the shadow of one rod thrown by the first light might be just in contact with that of the other rod thrown by the second light. By introducing coloured glasses in front of the lights, Rumford compared the illuminating powers of different sources with respect to light of a particular colour. The complementary tints exhibited by the shadows caused him to devise his theory of the harmony of complementary colours. One result is worthy of mention: it is a conclusion to which public attention has since been called in connection with "duplex" burners. Rumford found that with wax tapers the amount of light emitted per grain of wax consumed diminished with the diminution of the consumption, so that a small taper gave out only one-sixteenth as much light as an ordinary candle for the same consumption of wax. He says:-- "This result can be easily explained if we admit the hypothesis which supposes light to be analogous to sound.... The particles ... were so rapidly cooled ... that they had hardly time to shine one instant before they became too cold to be any longer visible." An argand lamp, when compared with a lamp having a flat wick, gave more light in the ratio of 100 to 85 for the same consumption of oil. One of the latest investigations of Rumford was that bearing on the effect of the width of the wheels on the draught of a carriage. To his own carriage, weighing, with its passengers, nearly a ton, he fitted a spring dynamometer by means of a set of pulleys attached to the under-carriage and the splinter-bar. He used three sets of wheels, respectively 1-3/4, 2-1/4, and 4 inches wide, and, introducing weights into the carriage to make up for the difference in the weights of the wheels, he found a very sensible diminution in the tractive force required as the width of the wheels was increased, and in a truly scientific spirit, despising the ridicule cast upon him, he persisted in riding about Paris in a carriage with four-inch tyres.