Parabiosis or Siamese-twinning is the formation of an organism consisting of more than one normal body. The additional anatomy may be only a section of a normal body, or it may consist of virtually complete bodies. Some define 'parabiosis' as the artificial joining of embryos; herein the term refers to embryos naturally conjoined due to a heritable mutation.
Parabiosis is a familiar phenomenon, observable even in our own species. It may be viewed as the failure of multiple embryos to differentiate fully. Thus while parabiosis creates a more complex organism, it is in a sense a reduction, in that it reflects a failure of normal embryonic differentiation, a truncation of the normal course of development. Such degeneration may reasonably be attributed to the proverbial "normally deleterious" random mutation.
Parabiosis is a crude mechanism. It is not capable of adding new parts within existing structural sets such as limbs, vertebrae, or auditory ossicles. This would require the extra embryos to be too finely positioned and too precisely reduced.
A difficulty for my model is that it requires the chaining of numerous bodies in a linear pattern, like the axial segments of segmented organisms, while observable instances of parabiosis seem more amorphous. I argue that such mutation was a reasonable possibility for an unsegmented organism in the ecosystem before the development of segmented species. Presumably many parabiotic configurations arose in this early period, including radially symmetrical organisms, while only a few have survived.
There is variation within populations of segmented species, including variation in the number of segments in a particular anatomical structure. Under the proposed model, such variation or "mutation" is part of the range of variability of a species; it is not parabiosis, and it is not a mechanism through which new segments are added in a evolutionary sense. Evolutionary reduction in number of parts is a long-term trend among populations, and is not negated by occasionanal atavisms.
Abnormally extensive segmentation is sometimes seen in deformities or in hybrids. Under the proposed model, these are not instances of parabiosis, but atavistic expressions of normally suppressed segmentation. Such segments are too nicely functional to have arisen through parabiosis or any other random variation. These phenomena can be explained as resulting from variation in a developmental program, where that program is a matter of the selective suppression of the expression of a fundamental pattern.
forward to Vertebrate and Arthropod Progenitors; Their Form and Origin