of malaria attributed to differences of soil were not really due
to other causes. I am far from saying that soil exerts no
influence on the endemicity ; but J can find no decisive evidence
that it does so, apart from the merely mechanical effect of
permeability. The matter deserves more exact study by modern
methods.
(3). Connection with slope. — Abrupt hillsides have little malaria
as a rule, but Anophelines often breed in dry beds of torrents,
as (for instance) described by F. Smith and A. Pearse in Sierra
Leone [1904]. I was infected in 1897 in such a place, Kalhutti,
near Ootacamund, India [February 1898], and the native servants
of the house were attacked there also. Small flat valleys among
I90 MALARIA IN THE COMMUNITY [Sect.
mountains are apt to be very malarious. Slope acts mechanically
on the drainage, but the labour and heat of climbing predispose
to chills and relapses.
Flat plains at the foot of mountains — called /erai in India —
are notoriously unhealthy. This is due to the fact that they
receive all the surface-drainage of the rain on the mountains,
the water flowing from which immediately stagnates on the
flat levels. After heavy rain a whole mountain seems to ooze
with moisture at its base, springs often appearing considerable
distances away on the plain.
Whether mosquitos often find their way far up slopes is
doubtful. In 1899 we observed scarcely a single Anopheline
in the barracks at Tower Hill, a small open hill in the centre
of Freetown, Sierra Leone, with numerous breeding-pools only
about 500 metres distant. Infections among people living at
a small height on slopes is probably more often due to the
visits they pay to the plain, rather than the visits paid by the
Anophelines of the plain to them.
(4). Connection ivith vegetation. — Several Anophelines are
known to breed in plants and trees. Certainly, many Culicines
do so, and also seem to like the shelter from sun and wind
given by dense vegetation. Our "moustiquiers" in Mauritius had
no difficulty in securing Anophelines in the densely - wooded
" river reserves." But I do not know any numerical researches
which have been made to prove that any malaria-bearing species
abounds more in woods than on open ground (see, however,
sections 57 and 63).
It has been proved statistically by Mr Walter of the
Mauritius Observatory that the damp exhaled by trees increases
the number of rainy days, and especially the afternoon rainfall
so frequently seen in the tropics. For this reason alone, there-
fore, trees should favour the breeding of mosquitos.
It is generally held that a screen of trees shuts out malaria
and mosquitos — this being one of King's original arguments
in favour of the mosquito theory. Stephens and Christophers
3o] CONNECTION WITH RAINFALL 191
also accept the hypothesis. It is a likely one, but better proof
is required.
The fact that the true Plasmodiidae have hitherto been
found only in men, monkeys, bats, squirrels and perching birds
is a curious one, which suggests an arboreal connection.
(5). Connection with rainfall is manifestly due, in the case
of summer rain, to the increased mosquito output. Rain also
tends to bring on relapses, and therefore to increase the factor —
since patients with frequent relapses tend to show more gametids
than those without them (section 20 (6)). Thirdly, I think that
it increases the biting factor, b, and quite possibly reduces the
recovery factor, r. Hence, on all counts it must tend to increase
the malaria.
Winter rain may possibly reduce r, but, if the winter is cold
enough, can have little other effect. Thus in Greece most of
the rain falls in the winter, when there is little new infection.
Much spring rain, however, has a very disastrous and well-
known effect, as it fills the pools just when the weather is
becoming warm enough for breeding. This has been especially
noted in Greece.
Statistics support the common statement that malaria
inoculation occurs most frequently at the beginning and the
end of great summer rains. At the height of these rains, when
they are copious, the ground is often covered with more or
less running water, frequently containing myriads of small
fish, and too disturbed for much breeding.
On the other hand, in countries with small total summer
rainfall, the maximum breeding is more likely to occur at the
height of it. There was a bad outbreak in the north of India
in 1908, when most of the rainfall was concentrated in the
month of August.
Innumerable statistics showing the connection between rain-
fall and malaria might be reproduced here. Unfortunately,
while they prove the existence of the connection, they do not
enable us to discriminate between inoculations and relapses.
192 MALARIA IN THE COMMUNITY [Sect.
Most probably the variation formula is affected by changes
in all the factors mentioned above.
(6). Connection with temperature, like that with rainfall, is
probably due to increase of all the factors and decrease of the
recovery factor r consequent on exposure to heat or sun.
Warmth is, of course, an essential to copious breeding, but
great dry heat must tend to desiccate the pools.
An important question remains to be considered. In low
latitudes the temperature is generally uniformly hot all the
year round, so that the Anophelines should be able to breed
at all seasons, especially where the rainfall is also fairly evenly
distributed. But in higher latitudes the breeding can occur
only during the short hot weather. Hence we should infer that
the Anopheline factor must always be higher in the former,
and the malaria more abundant. But statistics often show
that just the opposite happens. For example, in the north
of India, where the winters are sharp, the total fever rate is
generally considerably higher than in the south, where there
is no winter. Malaria is, or was, intense at Peshawar in the
extreme north, and scarce in Calcutta, Madras, Rangoon and
Colombo. But this law does not always hold, for malaria is
common in Panama, Colon, Lagos, Freetown and Port Louis
(Mauritius), all of which have a climate very similar to that
of the four Indian coast towns. (I write from personal
acquaintance).
The probable explanation is as follows. So far as we can
judge, not all the Anophelines can carry malaria, and various
carrying species differ in carrying power, thus modifying the
carrying factor, s. It does not follow that of all the Anophelines
of a country those with high carrying power breed best in a
uniformly warm climate. Stephens and Christophers showed
this well in their paper just referred to. Thus M, rossii, with
a low carrying power, prevails most in Calcutta and Madras.
See ( 1 1 ) below.
(7). Connection with altitude. — It is well known that malaria
3o] CONNECTION WITH ALTITUDE 193
tends to diminish and cease at an altitude of about 500-1500
metres above sea-level. The exact limit probably depends
upon the latitude. I was infected in 1897 at 1,800 metres in
the Nilgiri mountains in India. The disease abounds at
Cilaos, Reunion, at 1,214 metres. The text-books quote many
similar cases, but these often require verification. It is more
interesting to ascertain the gradual fall in the malaria curve
with altitude. Stephens and Christophers, misled, I think, by
insufficient random sampling, thought that altitude under 4,000
feet (1,219 metres) "does not seem to play an important part,"
but our copious figures of spleen rates in Mauritius (section 22)
prove that it does. That island, consisting almost everywhere
of plateaus sloping gradually downward and therefore being
capable of breeding at all the altitudes (from o to 549 metres),
gives an excellent opportunity for the enquiry, as will be seen
by study of the table referred to. The carrying Anopheline
was probably entirely M. costalis.
As is well known, the temperature of the air tends to fall
about 1° F. for every 300 feet of altitude (about 0"56' C. for
100 metres). The general decrease of malaria with altitude
is probably due mostly to this fall in temperature, but in
Mauritius, and perhaps in many other places, other factors
besides temperature may retard the breeding of the local
carriers at the higher levels. The following table, calculated
from that in section 22, gives the spleen rates and average
spleen for groups of altitudes : —
altitudes {feet) 0-300 3-600 6-900 9-1200 12-1500 15-1800
fall of temperature 0° F. i 2 3 4 5
spleen rates 44*5 41-2 31-5 8-5 10-2 47
average spleeft 2-98 3"o8 2-03 r23 r56 r22
The exceptional figures at 1,200- 1,500 feet were due to the
epidemic round Clairfond Marsh. Apart from these, the fall,
both in spleen rates and average spleen, is not noticeable under
600 feet, and then becomes very rapid up to 1,200 feet — after
which, under normal circumstances in Mauritius, the malaria
N
194 MALARIA IN THE COMMUNITY [Sect.
is slight, or possibly only imported. Other facts will be elicited
when governments get into the way of collecting spleen rates
annually, as they should do.
(8). Comparative freedom of centre of towns, — It is generally
stated as a commonplace that malaria abounds less in the
centre of towns and more in the suburbs ; and, from my own
general observation^ this appears to be the case in all the larger
towns known to me. But the matter requires strict enquiry by
measurements of malaria made on proper principles from the
centre outward.
The probable explanation is that the Anophelines, though
they may breed in small numbers in wells, cisterns, gutters,
waste from water-taps, etc., in towns, yet cannot generally
propagate so copiously and freely there as in the more open
regions in the outskirts. Probably, also, as the wealthier people
generally live in the better built and paved centre of a city, the
recovery factor is increased and the gametid factor decreased
there by the good medical treatment usually accessible to these
classes. But there is also the following factor.
(9). Effect of density of hwinan population. — Suppose that in
a locality the mosquito population remains the same, but
that the human population varies : what will be the effect of
this variation on the malaria ratio? By the static formula
M=i—4.oja; but a is the number of Anophelines, not in unit
of space, but per unit of human population. If, therefore, the
latter is doubled while the total mosquito population remains
constant, a will be halved ; and so on. Thus the static malaria
ratio tends to decrease with increase of the density of the
human population. That is, other things being equal and the
Anophelines being supposed to breed equally everywhere, the
malaria ratio should be higher amongst a scattered rural
population than in a dense urban one, because, evidently, the
number of Anophelines per person will be less in the latter.
But I am not sure that if the human population varies, the
mosquito population will generally remain the same. Unless
3o] CONNECTION WITH POPULATION 195
the latter find abundance of food independently of the former,
their numbers are likely to diminish if the former diminishes
(section 29 (7) ). This will depend largely upon whether the
local carrier is a domestic, sub-domestic or wild species, and on
other circumstances.
If the total mosquito population varies directly with the
human population, the factor a, and therefore the static malaria,
should remain constant — that is, change of density of the
human population will not affect the result.
If the mosquito population diminishes as the human popula-
tion increases, the malaria ratio should fall greatly — as in well-
drained towns. If the former increases with the latter but
more rapidly, the malaria should increase.
It may happen that when the human population begins to
increase the local breeding surface is already yielding its
maximum output of mosquitos. In this case the increase of
the human population should cause a decrease in the static
malaria ratio (section 29 (8) ).
In all these cases the malaria ratio is not, of course, the same
thing as the total number of patients.
If the local carrier belongs to a species or variety which
feeds almost entirely upon man, and if the human population
is greatly reduced, it may perhaps follow that this species can
no longer continue to thrive in the locality — so that the malaria
should die out. Dr Castel showed me a large marshy area in
Mauritius which, he said, was formerly thickly inhabited ; but
the disease became so prevalent there that the people deserted
it in large numbers. Now it contains only a few scattered huts,
the occupants of which show a low spleen rate (with P. costalis).
(10). The "■ regional factor" — It often happens that two
neighbouring tracts of country, apparently similar in all
respects such as climate, breeding surface, habits of people,
differ largely in the malaria ratio. Stephens and Christophers
[25th April 1902] attributed such variation to "undefined causes
which we have termed the regional factor. The regional factor
196 MALARIA IN THE COMMUNITY [Sect.
may be largely due to species, but more accurate and detailed
observations on the distribution of Anopheles and malaria are
necessary before this can be decided." It may be due to many
small things which, though not very apparent to the observer,
may largely affect the mosquito factors b, .y, a. Thus the local
carrier in the larval stage may require a certain kind of food
which abounds in water lying on certain soils present in one
locality and not in others. Again, certain soils may favour
special enemies of the larvae of the carrier — small fish, beetles,
cannibal mosquito larvae, parasites ; while certain classes of
vegetation may favour enemies of the adults. All these are
likely to be very potent causes of variation in the mosquito
factors, though they may not be easy to detect without long
enquiry. If we attribute the regional factor merely to "species"
of carrier, we have still to " explain our explanation," since we
must show why a given species abounds more in one area
than in another.
Hence I define as follows. By regional factors I mean all
those local conditions, apart from mere extent of breeding
surface, which influence any of the mosquito factors b, s, a\
that is, local conditions which influence the output per unit of
surface, the biting power, and the maturing power of the
local Anophelines.
(ii). The species factor. — This influences b and s, the biting
and maturing powers of the carriers, which probably differ
largely with the species or even variety of the carrier. Where
these factors are high, a lower number of Anophelines will
produce a given amount of malaria.
Thus tropical Africa appears on the whole to be more
malarious than India — that is, the chances of becoming infected
in unit of time and season in Africa are said to be greater in
Africa than in India. This may be because good carriers such
as P. cos talis and M. funesta abound more in the former. In
India the bad carrier, M. rossii, seems often to crowd out the
more pathophoric species.
3o] THE SOCIAL FACTOR 197
The local prevalence of good carriers or bad ones must
depend not only on regional factors but on the general
zoological laws which determine diffusion of species.^
(12). The social factor. — The factor b depends not only on
the appetite, energy and enterprise of the mosquitos, but also
on the intelligence, social status and habits of their victims.
Stupid, poor, lazy people, living in badly-made huts, without
much clothing and without mosquito-nets, are sure to be
bitten much more easily than more civilised races. People
who burn wood or cow-dung in their houses in the evening,
or who rub their skins with oil, earth or sandal-wood, or who
close their rooms at night, may perhaps be bitten less than
others. The subject is too complex for detailed examination
here. The habits of man and mosquito are probably often
correlated. Thus where the principal carrier is an out-of-doors
biter, people who sleep or work at night in the open are likely
to suffer. Alcoholism, opium, etc., lead to neglect of precautions.
Children are sure to be easy victims. Farm stock and dogs
may satisfy many insects, and punkas and fans drive away
others.
Neglect of precautions against being bitten is likely, not
only to increase the biting factor, but also, by section 29 (7),
the total number of mosquitos.
Famine, poverty and other diseases will reduce the recovery
factor.
(13). Possible el^ect of malaria on the Anophelines. — As earl)'
as 1898 I thought it possible that the parasites might injure
their insect hosts, as well as their human ones. If this happens
we can readily understand that an epidemic of malaria might
tend to limit itself by killing large numbers of the carriers
as well as men. Thus, during an epidemic year, so many of
the insects might die that the breeding might subsequently
be reduced for some time. But there are reasons against this
view. I could never satisfy myself that C. fatigans, even when
1 Possibly also on the insects' food (section 48).
198 MALARIA IN THE COMMUNITY [Sect.
extremely heavily infected with Proteosovia, died sooner than
when not infected at all. Probably they do so, but only to a
slight extent. Then again, only a small percentage of infected
mosquitos are heavily infected, so that the malaria infection
is not likely to make a material difference in their death-rate.
Lastly, even a severe epidemic among them will quickly be
compensated for by their rapid proliferation.
(14). Seasonal variation. — This is a matter of universal
observation. As a general rule in the northern hemisphere the
disease reaches its maximum prevalence in the autumn, say
October or November. At that point, a rapid decline, the
winter fall, generally begins, and this continues for several
months until early spring, say February or March, when the
winter minimum occurs. The spring rise now takes place, and
the disease tends to increase with more or less regularity until
the next autumn maximum. Of course, the seasons are reversed
in the southern hemisphere. Innumerable illustrative statistics
might be given ; but every one is acquainted with the
phenomenon, and the examples in section 20 will suffice for
our present purpose (note that in the Italian figures the
minimum is reached in June).
The causes of seasonal variation will be apparent from a
study of the variation formula given at the beginning of this
section. Thus the winter fall is probably due to diminution of
the factors h, .v, a, in consequence of the cooler temperature or
drying-up of the rains, or of both ; and also possibly to increase
of the recovery factor owing to the more bracing climate. On
the other hand, the spring rise is probably due to just the
opposite changes in these factors, and possibly also to the
emergence of Anophelines which have been hibernating through
the winter.
In the tables in section 20 I have given the ratios between
the average admissions for successive months. These can be
compared with the variation formula, by supposing that m is
the malaria ratio for any given month, and m^ the malaria ratio
3o] SEASONAL VARIATION 199
for the next month. Then, dividing the variation formula
throughout by ;//, we have,
injjn = I + b-sra{ \—in) — r.
Thus, take the monthly averages and ratios of the native
troops, and suppose that only the number of men actually
infected were admitted (section 31 (5)) into hospital every
month. Then, out of 43,330 men there were on the average
922 infected men in February and 1,030 in March. Thus m =
922/43,330 ; m^ = 1 ,030/43,330 ; m^jm =1-12; i-m = 0-98 ; and
therefore,
I • 1 2 = I + 0*98 b"sza — r.
Supposing that b, s, z, r have the values assigned at the
beginning of this section, we calculate from this that a =
(o*i2 + o'2)/o*oo49 = 65. That is, as a rough estimate, there
may have been about 65 carrying Anophelines per person
during the month February to March in order to increase the
infected persons from 922 to 1,030, and the admissions from
461 to 515.
Similarly, in order to cause the great rise in admissions
among the native troops from averages of 1,689 in September
^° 2,955 i" October, we may estimate roughly that 175
Anophelines may have been present. In order to cause the
commencement of the winter fall from 2,955 admissions in
October to 1,932 in November, we shall have from the formula
a negative number for a — that is, according to our data, there
should be no Anophelines biting at the time. But of course
our data are only speculative ones, and, as already stated, the
factors b and s, as well as the Anopheline factor, are likely to
vary with season. Moreover, my estimate that the recovery
factor r equals about 0*2 is probably too high for the earlier
months of infection.
(15). Anmial variation. — This also is a matter of general
observation. In every country there are " good years," " bad
years " and " years of epidemic." This variation, when it occurs
over large areas, is probably due principally to the influence of
200 MALARIA IN THE COMMUNITY [Sect.
climate, i.e., rainfall, number of rainy days and temperature, on
the Anopheline factor.
In the statistics of section 20, the numbers of admissions
differ not only for the whole years, but also for the same month
in different years, and for the European troops, native troops
and prisoners, respectively, during the same year.
It is interesting to note that a considerable rise in the total
annual admissions may, by our formulae, be produced by a
comparatively small rise in the Anopheline factor. Thus
among the native troops there were 11,293 admissions in 1903
against 19,567 in 1901. But by the static formulae the former
number would be given by « = 55 and the latter by a='jo. It
would be almost impossible, for the reasons given in section 29
(4), to detect this increase in the number of Anophelines (28%)
by ordinary observation. Hence the unwary observer might
assume that the rise in the number of cases was independent
of the rise in the number of Anophelines. In fact this is one
of the principal stumbling-blocks of students. They observe
a considerable annual variation in the number of cases with-
out, apparently , any great Anopheline variation. But I doubt
whether an Anopheline variation of as much as 50% or even 100%
would be easily detectable by the very unsatisfactory methods
of counting mosquitos at present known to us. If there are
over 40 different Anophelines to each person per month, we
should probably be able to find only two or three daily in his
house — and possibly less if the insects are " wild " ones. This
number might be doubled or trebled without producing a
change which would be readily noticed even after careful
observation, and long-continued random sampling would be
required to establish the variation over a tract of country. Yet,
as we have just calculated, a 28% increase in the Anophelines
might cause a 42% increase in static malaria — that is, an increase
of admissions from 11,293 to 19,567.
While, I think, the annual variation is generally due to
climatic causes, it is very possible that other factors often
3o] EFFECT OF IMPORTED CASES 201
influence it — such as poverty caused by increase in the cost
of food (which would diminish the recovery factor), or move-
ments of the population which might act in various ways.
(16). Effect of imported cases. — In section 28 I argued that,
whatever the original number of cases may be, the malaria
in a locality tends finally to settle down to a fixed static
ratio determined by the various constants. Thus if we
take the case of a village of 1,000 people, half of whom are
infected to begin with, with 60 Anophelines per person, the
rate would fall until, finally, about 333 persons are always
infected. If we had started in the same village with the
same number of Anophelines, but only one infected person,
then the rate would rise until the same number of infected
persons are found. But the reader must not infer that the
original malaria ratio is of no consequence — that it does not
matter whether many cases are imported or not. True, the
ultimate static ratio should be the same, but many months
may elapse before this limit is reached. If there have been
about 333 cases in the village for months or years past, and
we now add 167 imported cases, then (neglecting the small
change in the total population) we should have 500, 475, 455,
438 and 424 cases in the four following months; that is, 167,
142, 122, 105 and 91 more cases than there would have been