Let us now consider the discovery of Dulong and Petit, already alluded to. In 1819 they made the announcement that the atoms of simple substances, or elements, have equal capacity for heat. It must be explained that equal weights of different substances require different amounts of heat to raise them through the same interval of temperature. Thus, if the amount of heat required to raise the temperature of a gram of water from, let us say, o° C. to ioo° C. be taken as unity, it is found that only one-ninth of that amount is required to raise an equal weight of iron through the same range of temperature. Or, in other words, while the specific heat of water is i, that of iron is i, or in decimals, o. 112. The quantity of heat necessary to be imparted to one gram of water to raise its temperature through 1⁰ C. is termed a heat-unit, or calory; but sometimes the unit is chosen one hundred times as large, and represents the heat required for a rise of temperature from o° to ioo°; and the French make use of a unit one thousand times that of the smallest unit. Dulong and Petit's-^ discovery was, that if weights of the solid elements be taken proportional to their atomic weights, equal amounts of heat must be imparted to them in order to raise them through the same interval of temperature. The following table illustrates this fact, and exhibits some of the results obtained by Dulong and Petit:—

If the specific heat be taken as the heat required to raise the temperature of one gram of each of these substances through one degree, compared with that required for one gram of water, the atomic heats of bismuth, lead, gold, and the others represent the heats required for 208 grams of bismuth, 207 grams of lead, and so on. It is evident that they are all nearly equal. It should follow that the specific heat of solid hydrogen must be also 6, since the atomic weight is taken as 1.

These facts, though clearly indicating the numbers which should be taken for the atomic weights of the elements, were neglected, until renewed attention was called to them in 1858 by Cannizzaro, still Professor of Chemistry at Rome. He pointed out that all that can be gained from the analysis of a compound, for example an oxide, is the " equivalent" of the element. And as an element, such as iron, often forms more than one compound with other elements, let us say oxygen or chlorine, it therefore may possess more than one equivalent. But granting the atomic hypothesis, its atom can have only one definite weight. That atomic weight may, however, be inferred from its specific heat, or from the density of its gaseous compounds. Let us consider a concrete instance of each of these methods.

The analysis of two of the oxides of iron leads to the following results:—