View report page

32
Annual Report of the London County Council, 1913.
During 1913 the low level of typhoid fever prevalence now maintained for several years was
again observed and, indeed, as remarked in last year's report, in 1909, 1912 and 1913 there was
practically no autumnal increase at all. There are reasons for thinking that at any rate the large
majority of the cases of typhoid fever which occurred in London in those years, have been due to causes
other than the consumption of drinking water ; and yet if the number of times each individual must
be assumed to swallow single bacilli in the course of a year be multiplied by the figure representing
the population of London (four and a half millions), it transpires that the total number of instances
in which it cannot be reasonably inferred that the bacillus has not been consumed by Londoners
amounts to several millions.
But here, in addition to the question of dosage, it is necessary to bear in mind the
fact that although Dr. Houston has only happened to light upon two single bacilli indistinguishable
from typhoid bacilli, further examination may conceivably result in his coming
across samples which contain not merely a single typhoid bacillus, but a large number
of such bacilli. It should be realised that the total bulk of water examined, in the
course of the detailed investigations described in the Research Report, is almost infinitesimally small
in comparison with the amount of water consumed annually by the London population.
Apparently these examinations have related in all to only 29,470 selected colonies from 257 separate
raw river water samples (see Ninth Research Report, p. 6) or to a total amount of not more than some
twenty or thirty gallons of water. The sum of all the samples thus represents a proportion of only one
to many million parts of the total amount of water actually supplied in London during the period in
which Dr. Houston's examinations were made. To illustrate the danger of drawing conclusions from the
data thus available, a case which may perhaps be regarded as more or less parallel may be considered.
Two or three hundred children bear some such a like small proportion, to the child population of the
world, as that above referred to. No one would suggest that if examinationof 250 children, taken more
or less at random throughout the world, showed only two children presenting the rash of smallpox,
it could reasonably be inferred that smallpox was a negligible quantity at the present time so far as
the child population on the earth is concerned. There is, moreover, the consideration that in the case
of samples of water the limitations of the method of examination used are admittedly very great, so
that it would have to be assumed, in the more or less parallel case of the examination of the school
children, that while only two cases of smallpox were detected, there might have been in actual fact many
others. This consideration further emphasises the need for caution in interpreting the bacteriological
results just referred to.
There is a further point which should be mentioned. In Table 5 of the Tenth Research Report
there are two columns / and g, and a third column h, which gives figures corresponding to the ratios
0'
The averages of the numbers in columns / and g are given, but the average of the ratios in column h
is obtained by dividing the average of the figures in column / by the average of the figures in column g.
Such a method of calculating an average would be legitimate if it were desired, say, to ascertain the
mean death-rate of a number of different communities. If, for example, twelve towns were selected
with widely varying death-rates and populations, the true mean death-rate per unit of population of
the group ot towns would be rightly expressed by the ratio total deaths in the twelve towns, total population of the twelve towns,but if
on the other hand it were desired to ascertain the chance of death in a period of one year of an individual
residing for one month of the year in each of the twelve towns consecutively, this chance would be equal
to the arithmetical average of the annual death-rates per unit of the towns in question, and would
be independent of the size of the populations concerned. Thus, if it be assumed that the annual
death-rates per unit of each of the towns is a, b, c, d . . . I respectively then the chance of death
in one month to an individual residing in the first town for that period would be " , in the second
J -
etc., and his chance of death in the twelve months would be a+b+c+l/12 Having
regard to the fact that the latter instance is analogous to the case discussed by Dr.Houston it would appear
that the use of the arithmetical mean of the ratios f/g would have been preferable to the use of the average
upon which his deductions are based, for it is especially necessary to note that the variation of one
of the individual ratios in column h from the "average," depends in large measure on the time of
the year at which the sample is taken. Thus in Table 3 the figures for August and September are 185
and 640 and those for February and March 70,400 and 117,333, the average being 1,214. It is not
a little remarkable that the figures in the last column of this table present so marked a seasonal
variation; in August the degree of pollution of the filtered river water (as measured by the particular
test employed) was six or seven hundred times as great as in March and, if a seasonal curve of
pollution be constructed, it will be found that the water is much less polluted in the winter and
spring and becomes more polluted in the late summer and early autumn; the period of maximum
pollution, it may be observed, precedes by some six or eight weeks the time when typhoid fever is wont
to be specially prevalent.
It should be noted, however, that it is pointed out (p. 10) that the bad results obtained in August
are " probably due, in part at all events, to a slight multiplication of bacteria which can grow in bilesalt-agar,
but which are only very doubtfully related to sewage contamination ; and again, " the bilesalt-agar
test is an excellent one, but like all other tests it has its limitations," and ". . . any bacteriologist
of experiene will agree that under certain conditions, which do not readily lend themselves to definition,