Experiments were made on guinea pigs of (A), moderate sensitivity, and (B), high sensitivity to foreign protein of horse serum. Those of (A) were sensitized intraperitoneally with 0.01 cc. of horse serum 21 days previously; of (B) 84 days previously. In (A), a dose of 1.3 cc. of horse serum injected into the small intestine produced no anaphy-lactic shock, but the same dose injected into progressively shortened, ligated loops of small intestine produced shock in loops of 10 cm. or shorter. In a constant length of ligated loop, progressively larger doses (0.4-4.0 cc.) showed shock at dosage of 2 cc. or more. A shortened loop produced shock with a dose ineffective in the longer loop, and a non-effective dose for a given loop was made effective by increasing the volume of the dose with water. These results are explained as due to increased intra-intestinal pressure, which is held, therefore, to be the important factor in causing the anaphylactic symptoms in guinea pigs of moderate sensitivity. (B) reacted positively to a dose of 0.01 cc. horse serum injected intraperitoneally, whereas (A) required a dose of 0.1 cc. A dose of 0.4 cc. into the unligated intestine, or very small doses into extremely long ligated loops, produced shock symptoms in (B). The theory is advanced that minute amounts of whole protein are usually absorbed from the intestine and that highly sensitized animals react positively. Whenever intraintestinal pressure is raised this absorption may be so increased as to be detectable by less highly sensitized animals. Immunity to food proteins should therefore follow frequent ingestion: anaphylactic symptoms, infrequent ingestion.