Study of the morphological pattern of the arteries in the rotator cuff does not produce any evidence that the "critical zone" for ruptures and calcified deposits is much less vascularised than any other part of the tendinous cuff. This work, however, does show conclusively that the "critical zone" of the rotator cuff corresponds to the zone of the anastomoses between the osseous and tendinous vessels. Age does not seem to have a decisive influence on the morphology of the vascular pattern.

1 . The concept of the capsular mechanism of the shoulder joint with regard to recurrent anterior dislocation of the shoulder has been defined and a survey of the literature presented.

2. An anatomical, including an embryological, investigation of shoulder joints with special reference to the structure and function of the glenoid labrum and to the variations in the arrangement of the gleno-humeral ligaments and the synovial recesses of the anterior capsular mechanism is reported. The labrum, which is generally believed to be a consistent, fibrocartilaginous structure, is shown to be a redundant portion of capsular tissue and a continuation of the capsule as it attaches to the osseous glenoid rim. The fibrocartilaginous element is confined to a small transition zone at the capsular attachment in the great majority of cases. The great variability in the arrangement of the gleno-humeral ligaments and synovial recesses is stressed, and it is shown that an anterior pouch of variable size is present when the middle gleno-humeral ligament is attached to the scapular neck and not to the labrum.

3. The basic lesions of the anterior capsular mechanism found at operation for recurrent anterior dislocation of the shoulder in twenty-five consecutive cases using a modified Bankart procedure with a standard anterior approach to the joint are reported, and the findings are correlated with the results of the anatomical investigation. In most cases the lesions were found to be of the Bankart type with or without avulsion of the periosteum of the scapular neck. In four cases, however, the soft-tissue attachment to the anterior glenoid rim was intact; in those cases a large synovial pouch was present and the middle gleno-humeral ligament was either not discernible or it arose from the scapular neck. In all cases a postero-lateral notch on the humeral head was palpable and laxity of the subscapularis could be demonstrated. When measured, the joint capacity was always greatly augmented.

4. The present work shows, from a basic standpoint, that Bankart's original idea that the recurrent state was due to the failure of healing of the fractured fibrocartilaginous glenoid labrum is no longer tenable.

5. Finally, the anomalous attachment or the insufficient development of the middle gleno-humeral ligament in certain cases of recurrent anterior shoulder dislocation is shown to provide the anatomical basis for the recurrent state in these cases; this is the weak area in the antero-inferior part of the capsule which has been described in the literature for the past hundred years. Thus we have returned to the original view of Hippocrates.