an increase in the biting factor, due to the growth of the village
nearer to the marsh.
The disappearance of malaria from Britain has been ably
described by G. H. F. Nuttall, L. Cobbett and T. Strangeways-
Pigg [1901]. The observers made a careful study of the former
distribution of the disease in Britain, and of the present dis-
tribution of the Anophelines. The two coincided to some
extent ā that is, the malaria was within the Anopheline area,
but not so extensive. The disease, formerly severe, began to
decline early last century, but still lingered on until about 1850.
I was "clinical clerk" in St Bartholomew's Hospital in 1879
to a woman with greatly enlarged spleen and characteristic
fever, who lived in the Fen Country and had never been out
of England. Since that time endemic malaria has entirely
214 MALARIA IN THE COMMUNITY [Sect.
vanished from Britain, with the exception of two cases among
soldiers (footnote, section 14).^
The three British AnopheHnes are A. maculipennis, A.
bifurcatus and A. nigripes. They are all capable of carrying
malaria, but Theobald thinks that the first will not bite man
in Britain. The authors conclude that the disappearance of
malaria has been due {a) to reduction of AnopheHnes by
drainage ; {b) to reduction of population by emigration ; and
ic) to the use of quinine. I think that {h) is not sound. The
disappearance is almost certainly due to the reduction of the
Anopheline factor by drainage below the endemic limit, even
in the Fen Country, and also to enhancement of the recovery
factor by the more general use of quinine. I wonder, however,
whether there has been any change in climate, and also whether
glass windows have been used more since the beginning of
last century.
Many other examples might be cited, but the reader will
doubtless prefer to consider them for himself.
(22). Possible errors in the study of endejnicity. ā One is very
apt to find the local endemicity too high in comparison with
the local number of AnopheHnes to be found in houses. In
1897, in an intensely infected plantation near Ootacamund,
India, I found during several visits only one Anopheline (the
first I ever saw), although I searched the houses myself and
offered the people rewards ā and this was at the height of the
malaria season. In fact there were so few mosquitos of any
kind that I began to doubt the whole mosquito theory (just
before the discovery of the zygotes). The local carriers were
probably " wild " mosquitos which I did not know how to look
for; or else I may have missed the principal hatching-out
season. At Clairfond ten trained " moustiquiers " could procure
only a few P. costalis every night. We must never assume,
^ Sir Cliftord AUbutt tells me that he saw numerous cases in Cambridge in i860.
Nuttall, Cobbett and Strangeways-Pigg verified a case at Norwich in 1898 [1901,
P- 31]-
3o] SOURCES OF ERROR 215
without clear proof, that the number of insects seen or caught
during a few hours' or days' search affords any exact measure
of the number actually present during that period ā still less of
the number which may be present at other periods.
It is often observed that the malaria ratio is high where
there appears to be little breeding-surface ā that " dry " places
seem often to be more malarious than " wet " ones. But we
have no right to assume that the carriers always breed more
in the deep or permanent pools which constitute " water-
logging " than in the shallow, evanescent rain-pools which are
often the only waters found in the dryer areas. The latter
may perhaps give a much larger output at the proper season
than the former can yield all the year round (sections 29 (8)
and 30(1)).
In some cases the Anopheline factor is probably truly
below what we might expect from the malaria ratio ; but such
cases may be due to extraneous infection (19), or to immigra-
tion from neighbouring areas of high endemicity (16, 17).
Statements are frequently made to the effect that though
the number of Anophelines in a locality is large the malaria
ratio is small ā as, for instance, the case reported by Celli [1902].
Here again we must not assume that the number of Anophelines
actually found during a few days is an exact measure of those
existing all the year round. We have, perhaps, done the
counting just at the moment when the insects happened to
abound. Or, on comparing two localities, we may have been
concerned with a domestic species or variety in the locality
where the insects appeared to be more numerous, and a wild
species or variety in the other locality. Even where it is
proved by continued observation that the Anophelines do really
average a high figure where there is little malaria, this will
merely suggest that some of the other factors are unsuitable ;
for example, the local insects may belong to an immune strain.
Thus we have never succeeded in infecting the A. niaculipcnnis
round Liverpool from infected sailors.
2i6 MALARIA IN THE COMMUNITY [Sect.
In fact those who pretend to study the correlation between
the AnopheHne factor and the malaria rate are very prone
to make two mistakes : (a) they often fail to consider deeply
enough the principles involved, and (b) they are apt to base
startling " discoveries " on quite insufificient random-sampling ā
an error into which, according to Professor Karl Pearson,
medical men frequently fall. So far as the human reason
enables us to judge, there must be some correlation between
the two quantities. When, therefore, we are told ā as we
frequently are, especially in India ā that so-and-so has not
observed any such correlation, we naturally look for the
detailed figures of his investigations ; and when we find, in
place of these, mere statements of opinions and impressions,
we naturally infer that it is not the principles which are
worthless, but the observations. A person who thinks to
settle such difficult questions merely by running from village
to village, catching a few Anophelines and examining a few
people, can have little knowledge of the nature of scientific
evidence. To yield any worthy results at all, the enquiry
would demand the most careful and continuous study of all
the factors concerned (sections 29 (4), and 31).
A mistake frequently made is to suppose that there must
be some correlation between the number of Anophelines exist-
ing on a given day and the total percentage of infected persons.
But the former will tend to influence only the inoculation rate
of the future. The total percentage of infected persons will
depend upon past factors which may no longer exist. They
are often, or generally, connected, but not necessarily always so.
Equal errors may be made regarding the measurement of
human malaria. For example, the disease is often supposed
to be specially prevalent among soldiers, coolies or prisoners ;
but this may be merely a statistical error, due to the fact that
these classes are under more careful supervision than the general
public is.
{2^. I have attempted to examine as carefully as possible
3i] OTHER POSSIBLE FACTORS 217
the principal factors concerned in the spread of malaria, but
may have overlooked several of importance. The effect of
immunity, both among men and Anophelines, deserves much
closer study. So also do many questions connected with
difference of species of the parasites. It is by no means
certain that immunity against one species confers much or
any against another. If this is the case, many epidemics may
be caused merely by the entry of a species not hitherto pre-
valent in a locality ; and the simultaneous prevalance of all
three species may lead to a much more serious fever rate.
Koch observed [1900] that comparative immunity against
quartan seems to have little effect against the other parasites ;
but I am not sure that reaction against an existent invasion
in a person will have very much effect against a new infection
even by the same species.
The effect on the local malaria rate of desertion of marshy
areas (9) should be similar to that of drainage, and must not
be forgotten though it is not often witnessed. Conversely,
the effect of approximation to a marsh, as when people come
and settle near one (for example, the Clairfond outbreak just
described), should be similar to the effect of increase of marsh.
Obviously it will come to the same thing if we bring a marsh
close to human habitations, or bring human habitations close
to a marsh.
31. The Measurement of Malaria;ā In section 26 I gave
definitions of the terms malaria rate, endemicity, and inoculation
rate. We have now to consider how these and other estimates
can best be obtained for a given population inhabiting a given
area. Note to begin with that we can never obtain any such
rates exactly ; and also that the degree of approximation to
the truth must always depend on the amount of time and
labour we have to spare for the task.
Our estimates may be of two kinds, actual or comparative.
Comparative estimates are those which seek to compare the
2i8 MALARIA IN THE COMMUNITY [Sect.
proportion of infected persons in the same locality at different
times, or in different localities at the same time. It may
often be much easier to obtain comparative than actual
estimates.
Let us consider first the different possible methods, and
then the respective values of them.
(i). The parasite rate and index. ā By malaria rate or ratio
I mean the percentage or proportion of people who really
contain living plasmodia at some given moment ā not those
who show symptoms. By parasite rate I mean the same
thing. Now in section i8 (8) and (9) it was shown that
an average man contained about 3,000,000 c.mm. of blood,
and that a quarter of an hour is required to examine
1/50 of a c.mm. under the microscope. Thus in that time
we can search only 1/ 150,000,000th of a man's blood. In other
words, if there are less than 150,000,000 parasites in the
patient's blood the chances are that we may not find one,
even after a quarter of an hour's search. Hence we can never
hope to ascertain with certainty whether a single patient does
or does not contain plasmodia, and can therefore certainly
never discover the actual parasite rate of any number of people.
By parasite index I mean the percentage of persons in
whose blood plasmodia are found after fifteen minutes' ^ search
ā that is, of persons who contain more than about 50 plasmodia
to I c.mm. To ascertain even this is a laborious task, requiring
for 100 people twenty-five hours' continuous work.
Obviously the parasite index must always be less than
the parasite rate. What ratio will the latter bear to the
former? We do not know. Out of 100 infected persons
perhaps only 60 or 70 may contain more than 50 plasmodia
per I c.mm. of blood at a given moment. But this is a mere
guess, and no accurate studies have been made on the point.
The ratio certainly varies largely, but as we do not know it
with certainty, the parasite index gives no exact inform^ation
^ Some authors are satisfied with five or ten minutes' search.
3i] THE PARASITE INDEX 219
regarding the actual parasite rate. It gives valuable informa-
tion, however, on one point. Every person in whom parasites
are found is certainly infected. Hence the parasite index
gives at least the viinimuni possible percentage of infected
persons ā that is, we know that the actual parasite rate cannot
be lower tha^i the parasite index.
For comparative estimates the parasite index is still more
useful, because in them we deal only with ratios from time
to time or from place to place. Thus if it has varied from
say 20^ to 55^ in two large aggregates of persons examined,
we may infer that the actual parasite rate has varied to a
similar amount. We do not know what that actual parasite
rate was, with the parasite index either of 20% or of 55%,
but we conclude that its variation has been, ceteris paribus,
similar ā that is, from 20% to 55//.
As originally observed by Koch and Stephens and
Christophers, confirmed by Ziemann, Plehn, Annett, and
Button and many workers, the parasite index varies greatly
according to age, being generally highest in children (9).
Hence comparative estimates must be based on a study of
persons of the same age. Race, quinine treatment, and social
status seem also to have considerable effect.
The method is open to several sources of fallacy. The
skill of the observer may make a difference of say 20% or
more in the results obtained ; and the time employed in the
search for plasmodia is very important. In 1906 I obtained
a number of slides of blood from Greek children, and since
then these slides have frequently been given for study to the
classes of the Liverpool School of Tropical Medicine. On
every occasion that this has been done, one or more of the
students have discovered parasites in specimens in which they
had not been previously found ! In making exact comparative
estimates it is therefore necessary to employ observers of the
same skill, and to fix the time during which the search must
be made.
220 MALARIA IN THE COMMUNITY [Sect
But the principal weakness of the method is due to the
fact that, owing to the labour involved, only a comparatively
small number of people can be examined ā so that the error
of random sampling is likely to be very large. Thus, suppose
that we have examined 50 specimens for 15 minutes each,
costing 12-5 hours' work (exclusive of collecting and staining),
and have found plasmodia in 25 specimens, we have no right
to infer that the parasite index of all the people in the locality
will be 50%. The chances are, by Poisson's formula (8), that
if we examine another 50 persons we may find parasites in any
proportion of them from 30% to 70% ā the error being 20%.
Thus, if the same observer, using precisely the same methods,
examines 50 different persons on two occasions, and finds
25 infected on the first occasion and 35 infected on the second
occasion, he still has no right to assume that the actual
parasite rate {i.e., the malaria rate) of the locality has risen
between the two examinations.
The method is probably capable of much improvement by
quicker means of microscopic diagnosis (sections 65 (i) ).
(2). The spleen rate and index. ā This method has been in
use for a long time. We know that the spleen enlarges
sufficiently to be detected by palpation in a considerable
proportion of infected persons. Such enlargement is dis-
coverable with certainty and in a few seconds by the fingers
pressed under the ribs of the left side, and any one ā hospital
assistants, nurses and laity ā can detect it. The persons to
be examined are passed in a line before the examiner, while
another person records the results ; and with good manage-
ment 100 people can be thus inspected in an hour. Or else
we can do the work by house-to-house inspection. In section 22
I gave the results of a "spleen census" of 31,022 children in
Mauritius, and of 92,258 children in Ceylon ā so that very large
numbers of people can be studied by this method. It is open
to the following defects : ā
{a) The enlargement may be so slight in a small proportion
3i] THE SPLEEN INDEX 221
of cases, especially early infections, that it may
be overlooked in them if we use palpation only.
Generally, however, there is fever, or a history of
recent fever, in such cases.
{b) The spleen of healthy infants is sometimes so easily
palpable that the unskilled observer may think that
it is enlarged,
{c) Not all infected persons show palpable splenomegaly.
{d) Not all splenomegalous persons are necessarily infected.
{e) Other diseases, such as kala-azar and various anaemias,
cause splenomegaly (section 22) ; but the former is
limited to certain tracts of country, and the latter
are too scarce to affect large statistics. Generally
speaking, widespread splenomegaly is due to
malaria.
The advantages of the method are : ā
{a) That the enlargement can be detected, practically with
certainty and in a few seconds, by almost any one.
{b) That the method can be applied with little trouble to
enormous numbers of people, thus practically avoid-
ing the error of random sampling.
I have never known rupture of spleen to follow palpation ;
but the possibility of this must be remembered.
We must note that a smaller degree of splenic enlargement
can be detected by pe^xussion (as practised by medical men)
than merely by palpation. By the former method we can nearly
always detect some enlargement in malaria. Thayer [1898]
claims that it always occurs. Mannaberg says [1905] that out
of 132 cases of different types, "cachexias omitted," he found
it in all but one case ā in 15 cases (11*4^ of total) by per-
cussion alone, and in 116 (88%) by palpation also. Thus the
finer art of percussion added nearly one-eighth more to the
number of cases disclosed by palpation. Laveran agrees as to
the frequency of this symptom, but other authors give lower
222 MALARIA IN THE COMMUNITY [Sect.
percentages. It is probably often a question of care in
examination. Among sailors treated at the Royal Southern
Hospital of Liverpool, palpable enlargement is not very
frequent. Many writers give percentages without actuals ā
that is, nearly worthless figures. L. M. Hope [1904], in a
study of 1784 cases, microscopically verified at Pabna, Northern
Bengal, states that in 374, or 20%, the spleen was not palpable ;
but .she does not state whether the remaining cases were studied
by percussion. In 102 Greek children examined by me (4), the
spleen was palpable in 48, and parasites were found in 12 which
showed no enlargement (by palpation only). Probably, careful
percussion would have disclosed some enlargement in all of
these cases. On the whole, then, I think we should conclude
that some degree of splenic enlargement probably exists in
95-100% of all persons infected with malaria ; but that the
enlargement \s, great enough to be palpable only in about 75-90/^.
The reader should note this difference between enlargements
detectable by palpation or by percussion only. In public health
work, which demands the examination of large numbers of
people in order to avoid error of random sampling, careful
percussion is not generally practicable, or, at least, practised.
Thus by the term splenic index, I generally mean the per-
centage of persons in whom enlargement is detectable by
palpation only.
In order to determine the spleen rate accurately, it is
necessary to use careful percussion on all persons in whom no
enlargement has been found by palpation. If such persons are
numerous this may add considerably to the time required for
the investigation ā say two or three minutes for each person.
Another important point has to be considered. We have no
right to assume that the parasites are still present in all the
cases of enlarged spleen, especially in the older cases with large
chronic splenomegaly. Such cases may have become parasite-
free by establishment of natural immunity ; but we can never
know in how many the parasites have absolutely vanished, or
3i] COMBINED PARASITE AND SPLEEN INDEX 223
have only declined in numbers below the limit at which we can
readily find them.
In what proportion of persons with enlarged spleen are there
no Plasmodia at all ? It is impossible to say ; but we may give
20^ as a rough conjectural estimate.
Suppose, then, that 20% of persons with parasites have no
palpable enlargement of the spleen, and that 20% of those
with palpable enlargement of the spleen have no parasites ā or,
at any rate, that the two percentages, whatever they may be,
are about equal. It then follows that the palpable spleen rate,
or splenic index, will give a nearly exact measure of the per-
centage of infected persons ā that is, of the actual malaria rate
of the locality. But at present this theorem, which would be
one of great practical importance, is based only on conjectural
estimates.
Lastly, we must note that the spleen index is probably
modified considerably by local conditions, quinine, and the
age of the people examined (sections 22 and 31 (9)).
(3). Combined parasite index and spleen index. ā If we have
time we can use the parasite index and spleen index combined.
The people should be divided into four classes : ia) those
without enlarged spleen or detectable parasites ; {b) those with
both ; and {c) and id) those with one but not the other. Thus
out of 102 children examined in Greece, we found : ā
Moulki
Parasites
only.
5
Spleen
only.
23
Parasites
and spleen.
12
Nil.
22
Total
62
Mazi .
7
4
9
20
40
Total .12 27 21 42 102
Now if we use the term endemic index (section 26) to denote
the percentage of persons in whom malaria is diagnosed by a7ty
method, then the endemic index of Moulki at that time was
40/62=64-5%, and of Mazi was 20/40=50%. Provided that we
know that the splenomegaly is due only to malaria, and also
have time to work out the parasite index, this combined method
224 MALARIA IN THE COMMUNITY [Sect.
is obviously more exact than either method by itself. Thus
in both these villages together, the parasite index by itself
was only 33/102 = 32^, and the spleen index by itself only
48/102 = 47%; while the combined methods give a general
endemic index of 60/102=59%.
It is important to note that in this example, while the cases
with enlarged spleen numbered 48, those with parasites, but
without enlarged spleen, numbered 12, or 1/4 of the spleen
cases. If this proportion applies generally, the endemic index
will be found by adding 25% to the spleen cases. (This figure
is, however, too high by the results of Mannaberg and Hope
just quoted.)
Again, the cases with parasites numbered 33, while those
with spleen but without parasites numbered 27, or 9/ 11 of
the parasite cases. If this proportion holds generally, the
endemic index will be found by adding 82% to the parasite
cases.
It would thus seem that the spleen index is much nearer
the truth than the parasite index ā provided that the spleno-
megaly is really due only to the malaria.
Lastly, we observe that the figures both at Moulki and at
Mazi are too small to enable us to make exact comparative
estimates of the malaria rate of each village ; and we should
probably have done better if we had abandoned the parasite
index and had spent the time at our disposal in examining the
spleen of every person in the villages. Moulki, on the plain of
Kopais, contained 350 people, and Mazi, several hundred feet
above the plain, contained 575 people. The figures were
obtained in May to June [1906], at the beginning of the malaria
season.
The number of persons with parasites hit without spleno-
megaly seems to me an important figure, because such cases
are probably due to recent infections, before the spleen has had
time to become enlarged ā especially in children. There were
no less than 12/102=11-8% of such in these two villages,
3i] AVERAGE ENLARGED SPLEEN 225
suggesting that the spring infections had already commenced.
The weather was hot, and A. maculipennis plentiful.
(4). Average spleen and average enlatged spleen. ā By the
parasite index and spleen index, or by both combined, we
attempt to estimate the percentage of infected persons in a
locality ; but further information can be obtained by recording
the degree of enlargement found. For this purpose I suggested
in Mauritius [1908] a figure which I call the average spleen
(section 22). The observer notes whether the size of the spleen,
as roughly estimated by palpation, is unity, or about three times,
six times or nine times the normal size ā these sizes being
called normal, or small, medium or great enlargements. He
then multiplies the number of persons found to possess each
class of enlargement by i, 3, 6, or 9 as the case may be, adds
the products together, and divides by the total number of
persons examined. Thus in Mauritius we found, out of 30,137
children of fifteen years and under.
Normal spleens
Small enlargements
Medium enlargements
Great enlargements
Total
19,711
4,381
3,479
2,566
30.137
Thus the average spleen for children in the island works out at
2'54 times the normal size.
Another figure, which I call the average enlarged spleen, is
found in the same manner, but without considering the normal
spleens ; that is, multiply the small enlargements by 3, the
medium enlargements by 6, the great enlargements by 9, add
the results, and divide by the total number with enlarged
spleen. In the above example, the average enlarged spleen