(1) The object of a bomb calorimeter is to absorb the heat released from an accurately measured amount of chemical reaction and convert it into a temperature rise which can be measured with corresponding accuracy. Questions associated with the chemical reaction are often difficult, but will not be discussed here. This paper is concerned with the purely physical questions associated with the temperature rise. It is also concerned, in part, with the physical problem of producing a comparable temperature rise by introducing accurately measured Joule heat within the calorimeter. This procedure is necessary when the calorimeter is used for evaluating a thermochemical standard material, such as benzoic acid, with the intention of using it to measure the equivalents of other bomb calorimeters in which heat will be generated by combustion only. Reasons will be put forward for believing that aneroid bomb calorimeters are capable of higher accuracy than conventional calorimeters in which the heat is absorbed in well stirred water. It will be contended that aneroid calorimeters can be designed which are simpler, and quicker in use, than stirred-water calorimeters. (2) Apart from chemical considerations, the first requirement of a calorimeter is that its equivalent should be repeatable every time it is assembled, except for small variations in mass of substance, platinum, fuse and auxiliary material for which corrections can be made. The presence in the conventional calorimeter of water, whose heat of evaporation is very high, means that great care must be taken in design and assembly to minimize the evaporation and condensation of water, and to reproduce in every experiment any movements of water which are unavoidable. (3) The rate of transfer of heat Q between the calorimeter and its surroundings, during the period of the experiment, must obey an unchanging law such as Q. = k( T Tc) where T is the temperature of the thermometer in the calorimeter, and k and T c remain constant throughout the experiment. Corrections can be applied, for example, if Tin the equation differs for part of the time from the temperature of the thermometer; but errors are unavoidable if k and Tc do not remain constant. It is clearly an advantage if k is small, and this is best achieved by evacuating the space between the calorimeter and its outer jacket, and by lengthening the electrical leads which cross this space. Vacuum spaces are not usually possible round stirred-water calorimeters. As to the problem of keeping k constant, investigations by White l and others have shown that k is affected mainly by water vapour movement and by convection of air in the space surrounding the calorimeter. The