Davenport has reminded us that 'Epidemiological hypotheses must provide satisfactory explanations for all the known findings - not just for a convenient subset of them' (Davenport, 1977).

The current concept of the epidemiology of influenza A fails lamentably to pass Davenport's test. The common belief among doctors and their patients is that influenza virus, like measles virus, is surviving and causing epidemics in mankind by endless chains of direct transmissions of which each link in the chain is a person sick with influenza. Numerous observations and laboratory findings, some of which will be discussed, make the 'measles model' untenable for human influenza. The discovery that influenza A viruses parasitize many non-human hosts, both.

"Since the discovery of avian influenza it has become apparent how varied and complex are the modes of parasitism that have been evolved by influenza A viruses in numerous host species. Many circumstances help to determine the host parasite relationship. The species and the age. sex and nutritional state of the host are clearly important, as are the presence of other parasites or diseases, the exact serotype of the influenza virus, the previous influenzal experience of the host, and the geographical location and social interactions of the host species. "

"The classification of human influenza A viruses has undergone a series of changes since their discovery in 1933. and the latest nomenclature has not met with general approval. Classification depends in large measure on epidemiological findings and. conversely, it may itself shape our epidemiological concepts."

A subsequent section (pp. 17 22) attempts to assess how far along the road to speciation the human influenza A and B viruses have evolved, and to describe the natural history of the parasitic relationship with man.

"In chickens fowl plague may also cause heavy mortality ranging from 40 to 100%, but in this species mild epizootics also occur. Much light has been shed on the problem of pathogenicity by a study of the disease in chicken flocks in Pennsylvania. A mild influenza A outbreak occurred in April 1983 caused by H5N2 strains, but in the following October the same virus suddenly began killing large numbers of poultry"

Transplacental transmission does not necessarily cause death of the conceptus but may lead to symptomless latent infection of the offspring (Mensik. 1962). Epizootics of acute influenza in herds of swine latently infected with HswlNl virus may be being precipitated by an unidentified stimulus linked with seasonal changes (Easterday, 1975)."

"In the 1979-80 season some 20% of harbour seals (Phoca ritulina) on the north-east coast of the USA died of pneumonia caused by an avian influenza A virus closely related to that of fowl plague, a virus not previously found in mammals. "

(w.r.t Avian Flu in the wild...)"The acquisition of virulence was associated with changes at amino acid residue 23 near the stalk of the haemagglutinin and at amino acid residue 78 near antigenic site E (Webster. Kawaoka & Bean, 198(i)."

" Another avian strain previously found only in birds killed harbour seals off New England in 1982-3. Unlike any previous mammalian isolate, both these seal viruses replicated like avian viruses in the gut of chickens, so avian viruses appear to have caused natural outbreaks of influenza in seals (Webster et al. 1984). Swine influenza has been reported as spreading in turkey flocks (Andral et al. 1985)."

"Haemagglutinins analogous to human HI. H2 and H3 have been found in water birds, crows, chickens, bats, whales, squirrels, deer and lake water (Lvov & Zhdanov, 1983). The natural modes of parasitism of influenza A virus in non-human hosts therefore include at least the following: acute infection with and without illness, persistence, commensalism. vertical (transplacental) transmission and a variety of interspecies transmissions."

Whole virus persisted for only 30 days in chickens carrying influenza A virus, whereas HA and NA antigens persisted for GO days. Hydrocortisone administered on day 50 provoked the appearance of several strains unrelated to the original virus (Smolensky el al. 1978).

"A distinction needs to be made between persistence of virions and of viral antigens, the latter usually succeeding but occasionally preceding the former. Persistent infection seems to provide the best conditions for promoting variation of the virus. The word latency may need to be restricted to the persistence of subviral residues, retaining the term persistent infection for protracted infection with some sort of virion production. Chronic infection describes persistence with regular (even if scanty) production of infectious virions (Medvedeva. Aron & (iolubev. 1984)."

"The typo of cell in cell cultures, like the host species in animal infections, influences the result of infection by influenza A virus (Ahmed et al. 1981). Indeed, the host cell may select variants of influenza A virus and may be exerting as much evolutionary selective pressure as antibody (Schild et al. 1983); Xohinek. (ierhard & Schulze. 1985; Patterson & Oxford. 198(5)"

" Thus influenza B virus 14 R. E. HOPE-SIMPSON AND D. B. GOLUBEV seems already to have evolved into a separate species during its human parasitism (Alexander el al. 1981)"

"Strains isolated before 1940 were later considered to have belonged to a single major serotype whose minor variants were distinguished by this cross-protect ion. and the group was awarded subtype status, designated H0N1. Vaccination with any H0N1 strain conferred some protection against all the minor variants of the H0N1 major serotype.

In 1946 a novel major serotype, 'A prime', made its epidemic appearance, later designated H1N1 because its haemagglutinin differed from that in the H0N1 (old The epidemic process of influenza A 15 style) strains. The H0N1 (old strain) viruses promptly disappeared. and HlNl (old style) strains caused almost all the influenza A in the world population for the next 11 years, and previous infection or vaccination with H0X1 strains afforded no significant protection against them (Francis, Davenport & Hennessv. 1953)"

World dominance by HINI strains ended abruptly in 1957 when they were replaced worldwide in the pandemic caused by H2N2 influenza A viruses, a new subtype in which both surface antigens had changed. H2N2 viruses also had an era of 11 years of solo world prevalence.

The era of H2X2 influenza A viruses ended in 19(58 with their replacement worldwide by H3X2 strains in which only the haemagglutinin had changed significantly. The subtype of H3X2 viruses had near-solo dominance until 1977.

"These experiences led to a belief that during their era of prevalence strains belonging to one subtype somehow excluded strains belonging to all other subtypes of influenza A virus, though tolerating the presence of influenza B viruses. This apparent rule was broken in 1977 when H1N1 (old style) viruses reappeared after 20 years' absence, rapidly achieved world-wide distribution, and are still in 1986 oo-circulating with H3N2 viruses."

According to the doctrine of original antigenic sin ' the prevalence and magnitude of serum antibody against each individual influenza A strain are greatest in those who were young, and therefore probably experiencing their first attack of influenza, when the strain was circulating in the world and causing human epidemics' (Francis, 1960). Persons having suffered an HONl influenza A virus infection who were subsequently vaccinated with HlNl (old style) vaccine were found to produce an antibody response against HONl exceeding that against the HlNl vaccine strain. A problem in classification arose when it was later found that such anamnestic reactions of HONl and HlNl antibodies are not provoked by H2N2 virus infections. In relation to original antigenic sin there thus appeared to be two classes of influenza A viruses, one containing HswlNl, HONl and HlNl strains and the other containing H2N2 and H3N2 strains (Marine & Thomas. 1979). Not all observers, however, agree that H3N2 viruses fail to stimulate anamnestic antibody production in persons previously infected with HswlNl influenza A virus (Masurel, 1976). The breaching of the rule that successive major serotypes have eras of solo world dominance in 1977 when HlNl strains returned worldwide during an era of H3N2 prevalence suggested that solo dominance may operate only within each of the two classes as defined by original antigenic sin. HlNl strains may attack persons very recently infected with H3N2 virus (Frank, Taber & Wells, 1983)"

"Kilbourne's attractive hypothesis gains credibility by the natural behaviour of HswlNl and H3N2 influenza A viruses in swine and mankind (Shortridge, Cherry & Kendal, 1979; Ottis el al. 1982; Masurel el al. 1983). Valuable as the model is as possibly explaining the first origin of some major serotypes of influenza A virus in mankind, it does not explain the current behaviour of human influenza A. The pig or other host would need to be co-ordinating transmissions to man worldwide for each successive major serotype to appear worldwide within its first season. Human case-to-case spread from a single porcine source could not encompass the world population at such speed. Herald waves of influenza caused by the epidemic strain in the pre-epidemic season (Marine, McGowan & Thomas, 1976; Glezen, Couch & Six, 1982) only shift the problem back a year because the pre-epidemic wave would itself need to have been worldwide. The model also fails to explain the abrupt disappearance of the predecessor serotype, and of other features such as the ability to cause a large final epidemic as in January-April 1968, when most people had already been immunized by seven previous epidemics in the H2N2 era. The indiscriminate age distribution of those attacked in the final epidemic is also unexplained (Hope-Simpson, 1984). The Kilbourne model would be expected to produce a complicated pattern of simultaneous parasitism by varied influenza A viruses belonging to numerous major serotypes, rather than the ordered succession of clearly demarcated eras of prevalence, each consisting of a limited series of world epidemics from which only a few incommunicado communities altogether escape. Influenza A has survived in this way as a successful human parasite for at least a century, possibly for many centuries."

Records dating from before the discovery of human influenza virus in 1933 must be accepted with caution, but they sometimes describe influenza epidemics so excellently that they should not be altogether discounted. Shipboard outbreaks in the days of sail are of particular interest because the crews were often a$ sea for weeks or months. The fleet of Admiral Kempenfeldt sailed from Spithead on 2 May 1782, and remained at sea until influenza broke out with such intensity at the end of May that the ships were compelled to return to port early in June. Hirsch (1883) gives details of this and similar nautical outbreaks, and describes epidemics of influenza in remote communities after the rare arrival of a ship. 'The fact itself can hardly be doubted; while the striking thing appears to me to be that the strangers themselves, in all the cases, have remained exempt, or almost exempt, from the epidemic.' Such a phenomenon would be expected if the disease were being spread y symptomless carriers. A virologically confirmed example occurred at Point Barrow, Alaska, in 1935 when three healthy travellers arrived by plane from Canada, and 8 days later an epidemic of influenza began among the inhabitants (Burnet, 1945).

"It has been argued earlier in this paper that persistent infection is the mode of parasitism during which antigenic variation is most apt to occur, and the new concept accords with the laboratory methods used for selecting minor variants since the seminal observations of Archetti & Horsfall (1950) and Isaacs (1951). They showed that the Liverpool strain of HlNt (old style) influenza A virus grown m chicken embryo in the presence of Liverpool antibody produces a harvest of the Scandinavian variant, and vice versa."

"In some seasons more than one mutant of good evolutionary potential is produced and two or more may co-circulate. From two small brothers with influenza sharing the same bed in 1968, one yielded A/England/68 (H2N2) and the other A/ Tokyo/67 (H2N2) strain (Hope-Simpson, 1979). Circumstances in different localities sometimes favour one variant as against another, as in the great 1950-1 epidemic, when Scandinavian and Liverpool variants of H1N1 virus co-circulated in different parts of the world"

A new concept of the epidemic process of influenza A virus