Out of 180 cases collected by Brinton, 130, or 72 per cent., were scirrhus, thirty-two, or 18 per cent., medullary, and seventeen, or 9.4 per cent., colloid ; and upon the authority of these figures it has ever since been taught that ' scirrhus is found in threequarters of the cases of cancer of the stomach.' This statement is not only quite erroneous, but is so obviously based upon insufficient evidence that it seems almost incredible it could have remained unchallenged for such a length of time. In the first place, it will be observed that the number of cases is far too small to warrant any definite conclusion being drawn from them ; while the method of collecting isolated examples from various sources is open to the serious objection that it is only exceptional cases of a disease that are usually deemed worthy of publication. But, in addition to these initial sources of error, the growths themselves were classified, solely by their physical characters, the hard varieties being termed ' scirrhus,' and the soft ' medullary,' whereas it is now accepted that not only is it often impossible to determine the nature of the cancer by the naked eye, but even the microscope may fail to differentiate between the two varieties of the spheroidal-celled growth. Finally, Brinton appears to have been unaware of the existence of the cylindrical-celled carcinoma, which, according to Cornil and Ranvier, is the commonest of all; so that we must conclude that the cases of this description which occurred in his series were classified as scirrhus or medullary, according as the disease took the form of an infiltration or of a soft tumour.

The misconception that may arise from a restricted study of statistics is well shown by the fact that out of 1,348 examples of cancer of the stomach which we have collected and analysed according to the method employed by Brinton, 863, or 64 per cent., were described as 'soft," fungoid,' ' medullary,' or ' encephaloid ' growths ; 447, or 33 per cent., as ' hard ' or ' scirrhus,' and thirty-eight, or 2.9 per cent., as ' colloid.' These conclusions, while they agree closely with those arrived at by Welch, are diametrically opposed to the teaching of Brinton.

The only certain method of determining the character of a growth is by submitting it to a microscopical examination ; but, unfortunately, comparatively few of the recorded cases have been studied in this way. Perry and Shaw examined forty-four museum specimens of gastric cancer, and found that thirty-two belonged to the spheroidal-cell and twelve to the cylindricalcell type ; while out of 115 cases of our own, seventy-three, or 63.5 per cent., were described as spheroidal-celled, thirty-three, or 28.6 percent., as cylindrical-celled, and nine, or 7.8per cent., as exhibiting signs of colloid degeneration. It rnay also be observed that out of forty-one cases of spheroidal-celled carcinoma, it was especially noted that twenty-two were of the soft or medullary variety, and nineteen of the hard or scirrhous type. These figures appear to indicate that spheroidal-celled carcinomata are more than twice as common as the cylindrical-celled variety, and that colloid degeneration can frequently be detected by the microscope when invisible to the naked eye.

Situation

The symptoms of the disease depend so much upon its situation that the question of location is one of considerable importance. All writers are agreed that the pylorus is more often implicated than any other region of the stomach ; although whether the growth commences at the valve and spreads inwards, or develops near the orifice and becomes sharply limited by the valve, is a point which is not only impossible to decide, but one that is not endowed with any special value. According to Brinton, 60 per cent, of all gastric cancers are situated at the pylorus; Lebert's estimate was 59.6 per cent., Katzenellenbogen's 58.3 per cent., Luton's 57 per cent., Welch's 60.8 per cent., and Hahn's 36 per cent. In our own series of 265 cases the pylorus was primarily affected in 173, or 65.3 per cent.; while in 1,850 cases which we have collected from different sources, the disease was located in this region in 58 per cent.

Table 1.-An Analysis Of 1,850 Cases Of Carcinoma Of The Stomach, Showing The Frequency Of The Disease In Different Regions Of The Organ

Next to the pylorus in order of frequency comes the lesser curvature, with the contiguous parts of the two surfaces, which were the seat of the disease in 11.5 per cent. The cardiac orifice, with the tissues immediately around it, was affected in 9.8 per cent., while the fundus, which often escapes when the greater portion of the stomach is involved, was primarily attacked in only 1.5 per cent. The growth was limited to the posterior wall in about 5 per cent., to the anterior surface in 2.2 per cent, and to the great curvature in 2.8 per cent., while the whole or greater part of the viscus was implicated in 6 per cent. The stomach presented two or more separate tumours in 2.9 per cent. These facts indicate that in 79.4 per cent., or in about four-fifths of all cases, carcinoma commences in the comparatively small strip of tissue which extends from one orifice to the other along the upper margin of the stomach, and that its percentage incidence rapidly diminishes the further we proceed from the pyloric valve.