existence on any one day in the neighbourhood of such a
marsh of about 100 yards or metres square.
Of course the output is sure to vary from point to
point of a marsh, according to the breeding capacity of
different depths of water, etc. ; and it will also vary from
^ Nuttall and Shipley found equal numbers of males and females with A.
maculipennis also [January 1902].
i66 MALARIA IN THE COMMUNITY [Sect.
day to day according to the amount of rain, wind and
sunshine.
Correct and sufficient observations on these points are
much needed. They could be made by the use of small
square tents of muslin netting, supported by a central pole,
and pegged out with weighted margins over a unit of area
(say 1 square yard). We might thus ascertain not only the
mosquito output but the length of the aquatic life of mosquitos
under natural conditions, and other facts. Of course a number
of such tents must be used to avoid much error of random
sampling.
(2). The average life of the winged insects. — It used to be
thought that mosquitos feed on one occasion, lay their eggs
a few days later, and then die. But in 1898, observing that
this short period did not suffice to allow the malaria parasites
to develop fully in Culex fatigans, I ascertained that these
insects could be kept alive in captivity for a month, and
possibly much more, by repeated feedings [21st May 1898] ;
and I obtained the same results with certain Stegomyiae and
AnopJielines. Since then, similar observations have been made
on many mosquitos. Thus Goeldi [1905] kept . a female
Stegomyia calopus alive for 102 days, and a male one for
72 days. The average life in captivity of 15 females was
53 days, and of 1 1 males was 50 days. My general experi-
ence has been that it is easy to keep various species of
Anophelines alive in captivity for two or three weeks, but
not so easy to keep them longer. Nuttall and Shipley [1902]
kept A. niaculipennis for 56 days, and found that the females
tend to live longer as winter approaches, possibly in connec-
tion with hibernation (Woldert). It has long been known
that many mosquitos hibernate and aestivate. Females appear
to live in captivity longer than males. Better experiments
with large mosquito-houses, fixed in the open air under natural
conditions, should be made.
But such observations do not settle the important question.
NET FOR MEASURING THE OUTPUT OF MOSQUITOS FROM A MARSH— CLAIRFOND,
MAURITIUS.
BANKS OF A STREAM ROUGH-TRAINED FOR RS.O-37 A RUNNING FOOT FOR BOTH
BANKS (MAURITIUS). {^To face page i66.
29] AVERAGE LIFE OF MOSQUITOS 167
what is the average \\{g of mosquitos in nature? In captivity
they doubtless suffer from confinement, but, on the other hand,
are preserved from their natural enemies and from heat, wind
and weather, which probably destroy immense numbers of
them. We should like to know what percentages survive for
one, two, three . . . weeks. Numerous experiments on the
point could be performed, but have been neglected. In the
meantime, I have always taught the following hypothesis.
The average natural life of an animal is likely to exceed that
of any parasitic organisms which it may contain. The latter
have been instructed, so to speak, by the evolution of centuries,
as to the length of time they should spend over their development,
so as to have the best chance of being propagated to other hosts.
Thus Filaria bancrofti requires about three weeks to develop
in Culex. Now, if the average life of Culex were less than three
weeks, the filariae would have a much smaller chance of pro-
pagation. Doubtless, therefore, the average life of Culex exceeds
three weeks or a month. Similarly, plasmodia require about
ten days to develop in Anophelines, and I therefore suppose
that the average life of these insects reaches about three weeks
or more. But this reasoning suggests only the lower limit
of average life. Evidently a longer average life would improve
the chances of propagation of the parasites. The mosquito
death-rate must, of course, vary largely in consequence of many
factors, such as season, enemies, local conditions of shelter, etc.
On the whole, I think that we shall not be far wrong if we
accept our previous estimate that only about one-third of the
carrying Anophelines live for ten days — long enough to allow
the Plasmodia to mature in them.
(3). The proportion of mosquitos which succeed in biting human
beings. — With most species only the females, that is, let us say,
half the number, suck blood at all. When a number of females
are liberated all night within a mosquito-net occupied by a
man or by birds, only a variable proportion are found next
morning to be fed. Numerous observations have been made,
i68 MALARIA IN THE COMMUNITY [SECT.
by myself, among others, which show that the insects are not
very hungry for about twenty-four hours after hatching out
from the pupa, or immediately after laying their eggs. With
species which attack animals (birds, cattle, dogs, etc.), the
chances are that the proportion of insects which succeed in
biting men varies with the proportion which exists between
the number of men and of these animals in the locality. Men
must be more difficult to reach than defenceless animals (or
children). The richer classes often use mosquito - nets ; and
poor natives, especially Indians, generally cover themselves from
head to foot with a sheet during sleep, besides filling their huts
with smoke at night time, and keeping their doors and windows
shut (the common habit, even in hot climates). Access to people
in upper storeys, with small open windows, must often be
difficult to these feeble insects, especially while any breeze
is blowing.
I have suggested that possibly a quarter of the Anophelines
may succeed in biting human beings once, a third may live
for ten days more, and a quarter may succeed in biting again —
that is, that only about 1/24 of the females can ever have any
chance of carrying malaria. But this applies only to. the pro-
portion of mosquitos to each person in the place, and supposes,
moreover, that the human beings are evenly distributed in the
locality, and are fairly accessible. In thinly-inhabited places
the ratio will probably be much smaller, and in crowded ones,
larger. I think that many insects which have failed in pro-
curing a meal during the night may die of starvation (in my
experiments on birds a number of dead mosquitos were generally
found in the nets every morning). Numbers of gorged mosquitos
are probably devoured by bats, birds and spiders, or are killed
by sufferers. On the whole, then, an average ratio of 1/24 is
perhaps too high.
Many observers have studied the proportion of naturally-fed
Anophelines which contain plasmodia. In 1899 I found the
parasites in 27 out of 109 P. costalis caught in a military
29] MOSQUITO DENSITY 169
hospital in Freetown, Sierra Leone, in which one quarter of
the men examined contained them [1900]. Stephens and
Christophers found them in 6/70 of Anophelines in the same
town [5th April (?) 1900]; in 5-10% in Africa "as a rule";
and in some villages up to 50% [1903]. Ziemann, A. Plehn,
the Sergents, and many others have found similar rates.
Unfortunately, such observations do not help us much, because
the mosquitos caught in houses may possibly be only or mostly
those which have already obtained human food, and are waiting
for more.
In the huts of poor natives, and in badly-managed barracks
and hospitals where many unprotected people sleep in the same
room, a single mosquito may often be able to bite several
persons during the one night. In such houses the chances of
infection must be enormously increased, and the practice of
congregate sleeping must be one of the principal causes of the
diffusion of malaria.
(4). The number of Anophelines in unit of area. — Nothing
approaching accurate scientific work has been done on this
subject. There are many references in literature to " few
mosquitos," to " many mosquitos," to " swarms of mosquitos,"
to "des centaines de milliers de millions," and so on. Theobald
[1901, I. 72] describes having twice seen clouds of male and
female C. cantans in the English Fens, darkening the air, and
producing a sound which could at times be heard a quarter
of a mile distant. He says that W. W. Smith records that
a train in New Zealand " passed through a wall of mosquitos
three-quarters of a mile in length, twenty feet high, and eighteen
inches thick " ; and he mentions dense masses of gnats " like
columns of smoke." If there were ten mosquitos to the cubic
foot in this "wall," there would have been only 1,188,000 insects
in the whole collection — not a large number considering the
possible output (i). Such phenomena merely suggest that
occasional " swarming " occurs with mosquitos as with other
animals (including man).
I70 MALARIA IN THE COMMUNITY [SECT.
Anophelines hi houses have been frequently caught and
counted, especially by Stephens and Christophers and recent
observers in India. The numbers may vary from zero to many
hundreds in a single room, especially in thatched huts. They
may also vary from house to house, and according to distance
from breeding waters. It is impossible to quote any correct
averages. I have thought, as a general conjecture, that one
Anopheline for each human occupant might be adopted as
a kind of standard for comparison ; but, of course, near marshes
the numbers often rise to 50,100, or more in each room, or even
to each person.
We cannot estimate the number of any species in unit of
area by the number caught in houses in that area, unless we
know the proportions of that species found inside and outside
houses respectively, at the hour of the day when the search
is made. As we can never know this proportion exactly, the
number of insects caught inside houses is no exact guide to
the total number existing within the area of observation.
Different species appear to differ largely as to the amount
of time they spend in human habitations. I define do^nestic
mosquitos as those which pass a large part of their lives in
houses, such as C. fatigans, S. calopus, M. rossii ; though even
with these species we do not know exactly how many hours
they spend indoors and out-of-doors respectively. I define
sub-domestic mosquitos as those which enter houses only for the
purpose of feeding, and wild mosquitos as those which never
enter houses at all.^ Some species, such as M. mauritianus, are
found in verandas of houses, but not commonly in rooms.
The average output, average length of life, and average
number per unit of area of any species of mosquitos are
correlated quantities which can be ascertained only by the
most careful measurements, such as have never yet been
attempted (so far as I can ascertain).
1 Arribalzaga and Ficalbi give similar classifications. The latter's is based on the
habits of the larvae.
29] MOSQUITO DENSITY 171
In the previous section I computed roughly that malaria
is not likely to persist in a locality where the pathophoric
Anophelines number less than about forty different individuals
to each person during a month. It would be very difficult to
ascertain anywhere how many different mosquitos there are to
each person, but the attempt should be made. Mere personal
impressions on the point are apt to be very wrong. The
victim, surrounded by many assailants, tends to magnify their
number from one to ten, and from ten to a hundred. The
use of a white muslin hand net, with which his enemies can
easily be caught, will disclose the truth. I am inclined to
adopt the following standard. The mosquitos in a room
or veranda are nu^nerous if a single person is attacked by
more than five at a time, and very numerous if he is attacked
by more than ten. There are few houses in the tropics
where one is not solicited day and night by two or three
mosquitos.
If we count the number of Anophelines found in a set of
houses every day for a number of days, we could obtain the
average number to be found indoors for each person living in
the locality. But we cannot say that all these are different
mosquitos. We may, however, attempt a very rough com-
putation as follows : — Suppose that on the average there are
10 Anophelines to each person every day. Then, if all the
insects are changed every day — that is, if each insect lives only
24 hours on the average — there should be about 10x30 = 300
different insects to each person during a month of 30 days.
If, however, the insects live for 15 days on the average —
that is, are changed twice a month — there should be about
10x30/15 = 20 different insects to each person during the
month. If the insects live 30 days, there should be only
10 different ones to each person. Thus if n is the average
number of mosquitos found per diem per person, a is the
number of different mosquitos per person during a month of
30 days, and / is the average life of the mosquitos in days,
172 MALARIA IN THE COMMUNITY [Sect.
we should then have roughly a = jiXTplL If ^ = 27 and /= 20,
then a = ^o, the number roughly computed as the malaria-
maintaining limit.
To sum up — if we think over these points carefully (and
this has not always been done), we shall be convinced of the
great difficulty of forming any accurate notion of the mosquito-
density of any species anywhere. The enquiry would demand
a laborious study of the output, which must vary from week
to week ; of the average life, which must also vary ; and of the
average numbers found in houses and in mosquito traps
(section 12). It would require the services of a number of
trained " moustiquiers," and the error would amount to say
10% or more, even with the most careful observations.
(5). Variation of mosquito-density from place to place. — It
would be very interesting to determine the rate at which
the mosquito - density falls at different distances from a
single breeding-place. Stephens and Christophers [5th April
(.?) 1900] showed that in Freetown, Sierra Leone, the insects
were abundant in houses near certain breeding-waters, but
diminished markedly at greater distances. An exact enquiry
would be very difficult. We could estimate the numbers
caught in houses and traps at various distances from a single
pool in an otherwise sterile area. Here one source of error,
the relative domesticity of the species investigated, would
cancel out from the ratios ; and ceteris paribus the propor-
tions caught should indicate the proportions present. But
other sources of error, facilities for obtaining food and shelter
in different houses, might disturb the results. An empty
farm close to a marsh might, for instance, attract many fewer
insects than a small, crowded village half a mile away. Never-
theless, we know that as a broad general rule, mosquitos tend
to diminish with distance from their breeding-place ; but the
exact curve of diminution has not been ascertained.
(6). Variation of mosquito-density from time to time. — Here,
too, we could count the number of mosquitos in houses and
29] OUTPUT PER BREEDING-SURFACE 173
traps for a series of days, and the factor of relative domesticity
would be eliminated from the ratios. The total variation
obviously depends upon two factors, that of output and that
of longevity. If both the birth-rate and the death-rate are
increased or diminished, the total density might remain the
same, a fact which seems sometimes to have been forgotten.
(7). Variation of mosquito-density due to food, etc. — Abundance
of food certainly has a great effect on domestic Culicines,
which tend to swarm in crowded and poor habitations. Pro-
bably it has a similar effect on sub-domestic Anophelines,
though these may not be so much in evidence. I mean that
abundance of food probably tends to increase the output,
although the total breeding surface remains the same. This
would be due to the fact that the females find food more
easily and consequently lay more eggs. Thus malaria may
perhaps increase in a locality, not because of the increase of
breeding-places or rainfall, or because of the introduction of
more imported cases, but simply because an increase of the
human population has provided more food for the Anophelines
(section 30 (9), (12), and (21) ).
(8). Relation of mosquito-output to extent of breeding-surface.
— This is a point of great importance as regards prevention. If
we reduce the extent of breeding-surface to a given pro-
portion, what will be the exact effect on the number of
mosquitos ?
{a) The number of larvae in a collection of water will
depend (i) on the number of eggs laid in it, and
(2) on the suitability of the water — temperature,
shelter, absence of enemies, food.
{b') Probably that number cannot exceed a certain limit ;
that is to say, there must be a maximum possible
output per unit of breeding-surface at any season.
if) Probably also the actual output is often less than the
maximum possible output, the deficiency being
due to absence of enough food for the females,
174 MALARIA IN THE COMMUNITY [Sect.
destruction of the adults, inaccessibility of the
water, etc.
{d) The maximum possible output per unit of breeding-
surface is likely to vary with season, and to be
much greater during the warm season when the
food of the larvae is probably more abundant, and
their development quicker.
From these data we infer as follows : —
{e) Suppose that the breeding-surface of a locality is yielding
the maximum output, and is then suddenly reduced
in extent, say to half or a quarter the previous area.
Then a proportional fall must occur in the total
output of mosquitos in the locality.
(/") Next, suppose that the breeding-surface was yielding
only a fraction of the maximum possible output
when the extent of it was reduced. Then the fall
in the total output of mosquitos may not be so great.
The females, which formerly laid their eggs in the
part of the area which has been drained, may now
resort to the pools which are still allowed to remain,
and may increase the output of these by stocking
them with an additional number of eggs. Thus,
though the total breeding-area is reduced, the part
of it which remains may have a larger output ; so
that the total output in the locality may remain
the same as before.
{g) But this compensation has its limits. If the reduc-
tion of breeding-area has been great, the pools that
remain may often be inaccessible to the females,
or may become overstocked with eggs and larvae,
for which they cannot provide enough food. At
best, the total output cannot exceed the maximum
possible output of the waters which remain.
{h) If the drainage operations have been commenced early
in the season, before the breeding - season is fully
29] OUTPUT PER BREEDING-SURFACE 175
developed, and if they have not been complete, then,
as the breeding-season advances, the waters which
remain will still continue to have an increasing
output ; so that the total output of mosquitos in
the locality may continue to increase in spite of
the partial drainage operations, I say that, if any
breeding - waters at all are left, the output will
increase from month to month as the breeding-season
advances to its maximum ; but this does not mean
that the total output after the drainage operations
is not less than the total output in the previous
season before the operations. The thoughtless
observer, seeing an increase of mosquitos due to
the seasonal increase of output in the waters left
undrained, may jump to the conclusion that the
drainage has had no effect. But a comparison of
the total output before and after the drainage would
probably correct this error. The two issues have
frequently been confounded — as, I think, at Mian
Mir. Of course, if any breeding - waters at all
remain, a certain number of mosquitos will continue
to be poured out, especially at the height of the
breeding - season ; but it is scarcely likely, ceteris
paribus, that a small breeding-surface can have as
great an output as a large one, and it certainly
cannot have more.
We conclude then as follows : —
(i). That cete7'is paribus the output must tend to vary with
extent of breeding-surface.
(2). But that the two curves will not always exactly coincide.
For example, if the breeding-surface is reduced to a half,
then the output of mosquitos will also be probably reduced
very considerably ; but it may not be reduced exactly in the
same proportion. If the breeding- waters are entirely removed,
then, of course, the output in the locality must entirely cease.
176 MALARIA IN THE COMMUNITY [Sect.
(9). Flock-migrations of viosquitos. — We have hitherto con-
sidered the mosquito population of locaHties as if it consisted
solely of insects born in the locality. But, obviously, many
of the insects found in any area must have entered from without ;
and we must now examine the subject of mosquito-migration.
By flock-migration I mean the simultaneous movement of
large numbers of animals of the same species in the same
direction — such as we are familiar with in the case of wild
cattle, swallows, locusts, etc. Does anything of the kind occur
with mosquitos? Howard [1901] quotes a letter in which flock-
migrations of immense numbers of mosquitos are reported to
have been twice witnessed by the same observer in America,
It was supposed that the insects had originated in a large
marsh 35 miles distant ; and they were numerous enough to
cloud the sky, and bend down the grass with their weight,
Nuttall and Shipley conjecture [January 1902] that the
phenomenon may be due to " overstocking of a given locality
by a species." If it has occurred once, it ought to occur
frequently enough to be recorded more often. There is always
the danger that a large local hatch-out may be responsible for
the occurrence.
(10). Visitation of ships. — Mosquitos frequently visit ships
half a mile or more from the shore. We must not infer that
they have purposely travelled so far in search of food. Winged
animals which have once started on a flight across water seldom
have the sense to return. At Highcliffe, England, I once
watched numbers of butterflies {P. brassiccB) flying out to sea
from the shore on a still morning — none were coming back.
Birds and insects, lost in this manner and wearied with flight,
naturally board passing ships for rest. Sometimes, however,
a ship anchored close to shore may be attacked by mosquitos
which have perhaps scented their prey from a distance. On
the other hand, I have often been on board such ships without
noticing many mosquitos. Once I spent a night in a small
open boat, rowed down a river in Burma, but observed at the
29] TRANSPORTATION OF MOSQUITOS 177
time that not a single mosquito attacked us, though the night
was still and warm.
(li). Transportation by ships and vehicles. — I have known
Ctilex and Stegoniyia to breed in water-jugs and flower vases on
board ship, and they frequently breed in bilge-water and wooden
water barrels. But as soon as the ship starts on her voyage
most of the insects seem to be blown out of the cabins. Nuttall
[1899] says that Roe once observed a dozen foreign species of
mosquitos on board a ship lying at quarantine in New York.
This was probably a sailing ship, as the vibration of steamers
seems to prevent the insects coming to rest. Balfour notices
that Anophelines are brought into Khartoum in boats.
Every one who has lived in the tropics has observed that
mosquitos occur in carriages and railway trains. During the
heat of the day the insects seem unwilling to leave the vehicle,
though it is in motion, and may thus be transported consider-
able distances. But vehicles which introduce mosquitos may
also remove them, and it is absurd to suppose that the small
numbers carried in this way can often influence the malaria rate.
(12). Transportation by rivers and winds. — Nuttall, Cobbett
and Strangeways-Pigg [1901] suggested that rivers may trans-
port eggs, larvae and pupae, but I think that most of the
latter would be devoured before they have travelled very far.
Many people imagine that mosquitos are carried by winds.
Householders who breed them in their own premises like to
ascribe their presence to marshes some miles away. The
authors just quoted state [1901, p. 8] that Fernald described
how mosquitos at Cold Spring Harbour, Long Island, New
York, were blown there from a distance of 15 miles by the
south wind ; but Nuttall and Shipley [1902, p. 61] quote Weeks
as correcting this by showing the existence of many breeding-
places at Cold Spring Harbour itself. When I arrived at