1. The term “reinfection” has been defined as the secondary infection of a scarlet fever patient during hospitalization with Str. pyogenes belonging to a serologically different type from that producing the primary infection.
2. Of forty-seven scarlet fever patients nursed in a multiple-bed ward and swabbed twice weekly during their period of isolation, thirty-three (70.2 per cent) became reinfected with a serological type of Str. pyogenes different from that causing the primary disease.
3. In fifteen out of the thirty-three patients reinfected, the reinfection was “latent”, i.e. gave rise to no clinical signs, while in the remaining eighteen the reinfection was “manifest”, i.e. was accompanied by clinical signs or complications.
4. Patients nursed in cubicles or in a ward confined to infections with a single serological type did not show reinfection; their convalescence was progressive and there were no late complications.
5. The majority of complications occurring during the third week of hospitalization and subsequently, in multiple-bed wards devoted to scarlet fever, are due to reinfection.
6. Most reinfections occur during the third week in hospital at a time when patients are as a rule convalescent from their primary infection.
7. The most frequent mode of transmission of reinfection appears to be by direct contact of patient with patient.
8. Ten instances of “relapse” in scarlet fever (only three in the present series) are quoted; in all of them the patients were nursed in multiple-bed wards. In each instance the “relapse” coincided with the isolation of a fresh serological type of Str. pyogenes from the throat, and must therefore be regarded as a second attack of scarlet fever.
9. The various systems of nursing patients in isolation hospitals are discussed and it is suggested that scarlet fever patients should be cubicle-nursed if possible. Failing this they should be nursed by the bed-isolation method in multiple-bed wards. By setting aside small wards it might be possible to keep together patients who are all infected by the same serological type of Str. pyogenes; the number of such wards would vary with the number (usually three or four) of epidemic types current at the time.