In the formal theory of ferromagnetic anisotropy the magnetic energy of a crystal is expressed in terms of the magnetic field strength and the angles between the easy directions of magnetization and the field and magnetization vectors respectively. Although the magnetization curves of single crystals are in good agreement with this theory as regards the components of magnetization parallel to the field, this is not true of the transverse component. In this paper the treatment is considered with special reference to the transverse component for the planes (100), (110) and (111). According to the existing theory the transverse component should increase as the field strength diminishes whereas the experimental results show that it becomes zero in very weak fields. To account for this a simple treatment is here developed which covers the whole range of field strength. The directions of magnetization of domains are treated as distributed continuously in angle, rather than as restricted to a limited number of particular directions. By assuming that the proportion of the volume of domains magnetized in any direction is larger the lower the energy of magnetization in that direction, reasonable agreement with the experimental results is obtained. Although the distribution assumed is of the Boltzmann type, the energy of distribution is not of thermal origin and its value is very much greater than kT. It is likely that the distribution is due to internal stresses or lattice imperfections. The observations by Honda & Kaya have been newly corrected for demagnetization as regards magnitude and angle. The discussion is restricted by the errors in the experiment owing to demagnetization, non-uniformity of magnetization and preparation of the crystals.