In his report of 1893, Parsons goes into the subject of recurrent attacks in individuals in greater detail. He quotes several communications received from various physicians and health officers. These opinions differ, some believing that the disease predisposes to another attack; others, that there is no effect on the incidence in recurring epidemics; and still others believing that there is a small amount of acquired immunity. The communications are not based upon statistical evidence. He does find, however, an opportunity for statistical study in the industrial schools at Swinton near Manchester: “These schools were severely affected in March, 1890, 171 out of 589 children having suffered, or 29 per cent. In the first epidemic of 1891 they were again affected, but to a less extent, only 35 cases occurring. At that time there were in the schools 449 children who had been there at the time of the former epidemic. Of these 150 had had influenza in 1890 and 4 of them had it again, or 2.6 per cent.; 299 had escaped influenza in 1890 and 17 of these had it now, or 5.7 per cent. Thus, so far as these figures go, an attack of influenza confers a degree of protection which after the lapse of a year diminishes by one-half the liability to contract the disease.” Leichtenstern, like Parsons, recognizes the importance of distinguishing between relapses and recurrent cases. Relapses in influenza are not common. They usually occur after the patient is up, and about when he is ready to leave the house. These are not recurrent cases, but in the epidemics in the years following 1889 there were plenty of well substantiated cases of recurrent typical influenza in the same individual and some times even in entire families. During the 1889 epidemic, as during the 1918 epidemic it has been suggested by various observers that the apparent immunity among the very old was due to immunity developed as the result of previous epidemics, such as that of 1837, 1847 and 1857. Leichtenstern has collected the statistics from five different hospitals in which 8, 32, 35, 24, and 24 per cent. of individuals attacked in 1891–92 had already had the disease in 1889. Allbutt in 1905 remarked that whereas he had previously believed that immunity to influenza usually persists as long as six months, many cases had recently been brought to his notice where such an interval seemed improbable, where the succeeding attack was probably not a relapse but a new infection. He has seen two attacks apparently separate occurring in the same individual within two months. In the same year Moore wrote that influenza shows a decided tendency to relapse, a feature to which the indirect fatality of the disease is in great measure due. “So far from establishing immunity, an attack of this malady seems to render an individual more liable to contract the disease upon any future exposure to its contagion.”