The evidence here presented indicates that this special type of parotid tumor is both histologically and biologically similar to the thymic epithelial tumors. This is not surprising, if one considers that the parotid gland and thymus originate embryologically in proximity to one another in the primitive pharynx. In humans afflicted with Mikulicz's disease, the parotid gland attains huge proportions as a result of lymphoid accumulations. The term "benign lymphoepithelioma" has been applied to these enlargements (60). It is suggested, therefore, that the similarity between thymic tumors and the lymphoepithelial tumors of the parotid gland in mice is based on an active or a potential affinity for lymphoid cells, acquired by the thymus and parotid gland in the course of embryo-genesis. This potentiality is realized under normal conditions in the thymus gland, but only under exceptional and poorly understood circumstances in the parotid gland.

Inclusion of the hair-follicle tumors under the heading of epidermoid carcinomas (5) does not appear valid. The histology of these tumors has been studied closely (IS, 14) and it is certain that the invasiveness and the cellular characteristics of squamous carcinoma are lacking. These tumors cannot be placed in any previously established category. We have not seen metastasis, the origin of which could be traced to this lesion, among more than 100 animals bearing literally thousands of individual tumors of this type. These tumors should not be accepted as the equivalent of squamous carcinoma in any studies of pathogenesis.

The same restraints apply to the epithelial tumors involving the mandible and maxilla. The morphologic resemblance of some of these tumors to ameloblastomas has been previously noted (13), and their possible origin from epithelium of the dental lamina deserves further investigation. Squamitization occurs within some of these tumors, but they do not have other features of squamous carcinoma.

It is improbable that squamous carcinoma of the forestomach (6) is related causally to polyoma virus. We have encountered only one example of this among many hundred mice autopsied after inoculation with this agent (61). Histologically this example was similar to "spontaneous" carcinomas that occasionally occur at this site (62). It was readily transplanted by subcutaneous route.

The relationship of some primary lung tumors to polyoma-virus infection is problematic. Two primary lung tumors were listed by Stewart et al. (5), and Latarjet and de Jaco (43) reported several in mice receiving extracts from leukemic tissue. Stewart et al. (6) and later Dawe et al. (18) noted ballooning nuclei and cellular enlargement of bronchial epithelium, but whether these lesions can and do progress to tumors has not been established. That they well might is suggested by Rabson's observation (68) of multiple bronchial epithelial tumors in hamsters receiving large intratracheal doses of virus.

The pleural, pericardial, and peritoneal lesions described by Stewart et al. probably are related causally to the polyoma virus, and offer a unique opportunity for study of mesothelial proliferations. The absence of these lesions in our material (13) is perhaps explained by exclusion of noninbred Swiss mice from the study.

The tumors composed of mixtures of adipose cells and bone, and the adrenal medullary tumors arising in mice receiving this agent have been previously commented upon with regard to their novel character (IS, 14). The origin of the adrenal tumors from medullary cells is still open to some question because of their striking deviation from the histologic pattern of "spontaneous" medullary tumors, or pheochromocytoma (13, 21). They have also failed to show any functional activity.

Several other tumors have occurred too infrequently in mice inoculated with polyoma virus to warrant an opinion as to whether they were spontaneous or arose as a result of treatment. It is presently largely a matter of opinion as to whether the harderian-gland tumors (6,13), the urethral-gland tumors (18), the subcutaneous hemangioendotheliomas, and the hemangiomas of the liver (5) are specifically related to the activities of the agent.

A peculiarity of some small tumors of the extraorbital lacrimal gland of the mouse deserves brief description here because it raises some perplexing questions. This peculiarity consists of greatly enlarged cells in the acini immediately adjacent to small foci of tumor (figs. 3 and 4). Both cytoplasmic and nuclear volume are increased several fold. Moreover, mitoses in these cells are very numerous, even though the cells are clearly not part of the tumors as they are ordinarily recognized. The mitotic figures also contain numbers of chromosomes that place these cells definitely in the polyploid category (fig. 5), and frequently one or more chromosomes appear to have been separated from the spindle in a manner not unlike that seen after radiation injury. Does this phenomenon represent a different, non-neoplastic response to the virus? Is it a response not to viral infection, but to some product of cell-virus interaction released by the adjacent tumor cells [see Stanton (14) and Morgan (64)], or is it a response related only to nonspecific necrosis of other parenchymal cells, similar to the phenomenon described by Teir (65) in extraorbital lacrimal glands of rats inoculated with extracts of normal lacrimal-gland tissue?

In addition to the neoplastic response to polyoma virus, several nonneoplastic or possibly preneoplastic responses by the mouse deserve mention. These include hypertrophy and hyperplasia of epithelium of the renal tubules, the "runting phenomenon" first described by Stewart et al., and the recently reported occurrence of hydrocephaly in mice inoculated intracerebrally during the suckling period (66). The renal tubular lesion contains elements both of necrosis and of proliferation, and has been classified as preneoplastic by Buffet et al. (57). This lesion offers a challenge to investigators interested in cocarcinogens and endocrine influences. Would a steroid influence shift the balance in the direction of neoplasia at this site, as suggested by the influence of estrogens on the development of renal tumors in hamsters (67)?

In the opposite direction, is it possible that a necrotic response on the part of mesenchymal cells, similar to that observed in tissue cultures, could account for retarded bone growth and the consequent runting phenomenon? Studies of bone growth in infected organ cultures, after the method of Fell (68), might prove helpful in answering this question. It is anticipated that the viral studies to be reported later in this symposium by Dr. Rowe (80), will also bear on this point. So little is known of the pathologic anatomy of the hydrocephalic phenomenon associated with this virus that it would be premature to make speculations as to mechanism at this time.

The hamster, rat, and Mastomys respond to polyoma virus in what appear (deceivingly?) to be less complex patterns. Among these animals, only one type of epithelial tumor has thus far been demonstrated, and this in the golden hamster. This finding was recently made by Rabson et al. (68) who observed multiple proliferations of the bronchial epithelium in hamsters exposed to heavy doses of virus by intratracheal insufflation. The lungs of these animals were studded with proliferating cell aggregates, apparently originating chiefly in the smaller bronchi and extending into pulmonary alveoli, where they presented an appearance similar to the pulmonary lesions of squirrels receiving the squirrel fibroma virus intratracheally (69). It is of special interest that occasional foci within some of the nodules also showed squamous characteristics.

Mesenchymal tumors of at least 2 types-angioid, and solid spindle-cell- have been seen in hamsters (22, 27-29), rats (6", 15), and Mastomys (26), after subcutaneous introduction of virus. The situation is made complex by the fact that in each of these species the anatomic distribution of the tumors is different. There are already differences of opinion as to whether the vascular lesions of the liver in hamsters represent neoplasia in even a broad definition (70). In the hamster, subcutaneous tumors may be either solid fibromas or fibrosarcomas, or angiosarcomas (29). Primary tumors in the liver and lung of hamsters have been classed as angiomas or angiosarcomas (22). Tumors of the heart, large vessels, kidneys, and serosa of the gut present the picture of solid sarcomas (22), as do intracranial tumors occurring after intracranial inoculation (71, 72). In the rat, tumors are not so widespread in the mesenchymal tissues. The renal tumors have either a solid or angioid spindle-cell make-up, and the subcutaneous tumors are solid fibromas or sarcomas (6). Those of the skull (after intracranial inoculation) are bone-forming (16). Mastomys responds to subcutaneous inoculation in a manner somewhat intermediate between hamster and rat, with solid fibrous tumors of the kidney, angioid tumors of the liver, and subcutaneous fibrous tumors (26). It is interesting that solid fibrous tumors of the kidney and subcutaneous fibrous tumors are the only types common to all known responsive animals. This suggests that mesenchyme of the renal medulla and of the subcutaneous tissues may be less "species specific" than other tissues. In this connection, it should be noted that the renal site happens to be an opportune one for investigation of possible relationships between morphogenetic influences, viral susceptibility, and type of response. The first step would be to determine if these tumors arise from the metanephric mesenchyme, as is suggested by their location at the corticomedullary junction. A technique for isolating metanephrogenic mesenchyme and epithelium of the ureteral bud has been made available by Grobstein (78, 74). It is fascinating, but too early, to speculate far on the implications of Grob-stein's observation (78) that epithelium of the salivary gland (the organ most responsive to polyoma virus in the mouse) is the only epithelium, except that of the ureteral bud, which is capable of inducing tubule formation in the metanephrogenic mesenchyme, in tissue culture. This set of circumstances may be purely coincidental, as it was shown also by Grobstein (73) that spinal cord, even from the chick, also has a strong inductive action on metanephrogenic mesenchyme of the mouse.