A transdomain remanent magnetization is defined as any RM that is acquired in association with a change in the number of domains in a sample; consequently, this means that the sample must have undergone domain wall nucleation or denucleation. Moon and Merrill [1986b] first defined the term and carried out calculations to suggest there might be a very stable transdomain viscous RM (VRM) that occurs during superchrons. The calculations used for this prediction were one-dimensional, and to obtain a significant transdomain VRM they arbitrarily lowered the activation energies between LEM states by more than an order of magnitude. Although one expected that more accurate modeling would lead to lower activation energies---and this expectation has been realized---the reduction is far too small to allow for the occurrence of a significant transdomain VRM [ Dunlop et al., 1994]. There should be little doubt that transdomain RM can occur in other situations; for example, it should be clear that a change in grain volume (either an increase or decrease) will often lead to a change in the number of domains present, and any RM acquired during this process would be called a transdomain RM. Hence it is almost certain that on occasions a transdomain chemical RM (CRM) occurs. Ye and Merrill [1995b] argue that although transdomain processes may occur in the case of TRM, they will not be common. Nevertheless, there is good observational evidence that domain wall denucleation occurs in some titanomagnetites during cooling, an indication that at least on occasion some transdomain TRM occurs [ Halgedahl, 1991]. All the above conclusions that have been derived from theory need to be tempered somewhat because factors such as stress have not yet been properly included in the calculations.
One of the most surprising experimental results of the past few years is shown in m [ Halgedahl, 1991]. Similar results were obtained by Halgedahl for other grains. Repeated heating and cooling of this sample produced LEM states that ranged from one to eight domains. Ye and Merrill [1995b] argue that this broad range of states occurs because of thermal fluctuations just below the Neel temperature. At very high temperatures (i.e., close to the Neel or Curie temperature), the conventional picture of domains breaks down because the various anisotropy energies, such as magnetocrystalline and magnetostriction anisotropy, become insignificant compared to the exchange energy [e.g., Moskowitz, 1993]. Even closer to the Neel temperature, the thermal fluctuations cause the magnetic spins to form clusters. To treat this problem, Ye and Merrill [1995b] use a renormalization group theory method that effectively ignores everything but the exchange energy and thermal fluctuations; the method is only strictly valid within a degree of the Neel temperature. Regardless of whether the details of this method are supported by future work, they point out that if thermal fluctuations close to the Neel temperature are responsible for the wide range of observed LEM states, then there may be important consequences for paleomagnetism. They predict that grains in rocks containing secondary RM will often exhibit a significantly different range of domains from ``identical'' grains in rocks that have only a primary RM. For example, they predict that the range of observed LEM states will typically be much smaller in a rock containing a grain growth CRM than in a rock containing a TRM. In particular, this means that domain observations in the future may play an important role in distinguishing rocks that have been remagnetized.