= CO
EOO
•CM
ICO

ILO

'CD

CO

Gt. Brit, Ministry of Health
Bacteriological studies

REPORTS

ON

PUBLIC HEALTH AND
MEDICAL SUBJECTS.

No. 18.

BACTERIOLOGICAL STUDIES :

1. The Influence of Immune Serum on the Biological
Properties of Pneumococci. By F. Griffith, M.B.

2. Bacterial Variation and Transmissible Autolysis. By
A. Eastwood, M.D.

MINISTRY OF HEALTH.

LONDON:
PUBLISHED BY HIS MAJESTY'S STATIONERY OFFICE.

1923.

Price is. od. Net.

REPORTS

ON

PUBLIC HEALTH AND
MEDICAL SUBJECTS.

No. 18.

BACTERIOLOGICAL STUDIES :

1. The Influence of Immune Serum on the Biological
Properties of Pneumococci. By F. Griffith, M.B.

2. Bacterial Variation and Transmissible Autolysis. By
A. Eastwood, M.D.

MINISTRY OF HEALTH.

LONDON:
PUBLISHED BY HIS MAJESTY'S STATIONERY OFFICE.

1923.

Price is. od. Net.

SI

DC! ^
863941-

Ill

PREFATORY NOTE BY THE CHIEF MEDICAL
OFFICER.

To The Right Hon. Neville Chamberlain, M.P.,
Minister of Health.

Sir,

1. I beg to submit two reports from the Ministry's Patho-
logical Laboratory which form additional contributions to the
study of epidemiological problems by bacteriological methods.

2. Dr. Griffith has discovered a new and simple way to
distinguish colonies of virulent and non-virulent pneumococci,
and has shown that strains derived from such different colonies
differ not only in virulence but in their antigenic characters.
This observation is of great importance in the choice of strains for
the preparation of protective and therapeutic sera, since sera
prepared with strains from the non-virulent colony have almost
no protective action against infections with the virulent strain.

Further, Dr. Griffith has found that exposure of a virulent strain
to the action of a specific immune serum invariably leads to the
appearance of non-virulent colonies, and, if the exposure is
prolonged, to the transformation of the virulent pneumococci
into an entirely non-virulent strain. This fact suggests that
here we have one, at least, of the ways in which immune serum
exerts its protective action in the animal body.

3. These observations are of wider interest, however, as a con-
tribution to the study of bacterial variation. In recent years
much attention has been given by bacteriologists to elucidation
of causes which determine variations in the virulence of bacteria.
The disconcerting fact has been brought to light that a pure
culture, bred from a single colony or from a single bacterium,
is not necessarily a collection of homogeneous individuals; it is
often an aggregation of units which, though all true to species,
differ from each other in certain biological respects amongst which
is the property termed " virulence." It has been shown that
no simple explanation of this phenomenon is possible owing
to the many influences concerned.

4. Dr. Eastwood has endeavoured to extricate some general
principle from the confusing mass of data which have accumulated
around the subject of bacterial variation. He shows that these
observations of Dr. Griffith can be brought into line with other

* (78)20179(19000) Wt 5051— 529a/296 750 7/23 E A S (T.S. 3774) A 2

recent work on bacterial variants, in chiding the much discussed
" Twort d'Herelle phenomenon " of transmissible bacterial
autolysis. He points out that all these different manifestations
of bacterial changes have one factor in common; the changes
only occur in bacteria which are actually growing. He therefore
concludes that they must be due to some interference with vital
processes at the nascent stage of bacterial growth. On this
assumption he proceeds to formulate a working hypothesis
which may help to elucidate general principles. The constituents
of living bacterial protoplasm have two functions (1) Catalytic
i.e., the preparation of food material with the aid of their
appropriate enzymes; (2) Synthetic, i.e., the building up of new
protoplasm. Any disturbance of the delicately adjusted balance
between these two activities, whether by change of environment,
pabulum or otherwise, must tend to the production of variants. In
his report Dr. Eastwood furnishes a lucid account of the known
facts concerning bacterial variants and discusses them in relation
to the interesting hypothesis which he has formulated.

I have the honour to be,
Sir,
Your obedient servant,

GEORGE NEWMAN.
Whitehall,

June, 1923.

BACTERIOLOGICAL STUDIES.

PAGE

1. The Influence of Immune Serum on the Biological Pro-

perties of Pneumococci. By F. Griffith, M.B. • 1

2. Bacterial Variation and Transmissible Autolysis. By

A. Eastwood, M.D. 14

* 10179

L— THE INFLUENCE OF IMMUNE SERUM ON THB
BIOLOGICAL PROPERTIES OF PNEUMOCOCCAL

By Fred. Griffith, M.B.

More exact knowledge of the natural variations in virulence
and antigenic qualities which pathogenic bacteria may undergo
is needed for two main purposes. In the study of epidemiological
problems this information forms a necessary part of the data
which should be established ; and for the purpose of serum and
vaccine therapy it is equally important to know in what respects
immunological characters may be influenced by transfer of the
bacterium from the animal body to the test tube and by culture
on different media.

The conception of a " pure culture " of a bacterium as a
number of absolutely identical individuals is no longer tenable.
Arkwright,* in a recent interesting and suggestive article on the
subject of variation in bacteria, with particular reference to
B. dysenterice (Shiga), B. typhosus, B. paratyphosus B. and
B. enteritidis, has shown that, in all the cultures he has worked
with, two forms almost invariably make their appearance under
certain conditions. These he has designated R (rough), and
S (smooth), owing to the distinctive characters of their colonies
on agar plates. Further important differences between the two
are that each form appears to have a different main antigen, both
of which are present in the parent stock; and the R form, unlike
the S, does not form a stable emulsion in 0*85 per cent, solution
of sodium chloride. In the bacterial species upon which Ark-
wright reports the normal, at any rate superficially, resembles
more nearly the S form, the R form being the " variant."

Beniansf had previously recorded similar changes in a culture
of dysentery (Shiga) bacilli.

Mary CowanJ has described similar varieties of colonies in
cultures of streptococci; she states that "individual strains of
streptococci can be ' dissociated ' into ' rough ' and ' smooth '
types which show remarkable differences in virulence for labora-
tory animals."

Andrewes,§ in his investigation of the serological types in
the Salmonella group, has also obtained two varieties of colonies
from the same strain. In this case the colonies are identical in
appearance but show differences in their immunological reactions ;
in the one form the antigen which evokes and reacts with the
specific agglutinins is dominant, whereas the capacity to produce
and react with the group -agglutinins is dominant in the other.
Although Andrewes finds that the colonies of the two types are

* Journ. Path, and Bact., XXIV., p. 36, 1921.
f Journ. Path, and Bact., XXIII., p. 171, 1920.
X British Journ. Exper. Path., III., p. 187, 1922.
§ Journ. Path, and Bact., XXV., p. 505, 1922.

£ 2

indistinguishable in appearance, there seems to me little doubt
that both he and Arkwright are dealing with essentially the
same phenomenon and that the differences in some of the results
of the two workers merely represent different stages in the
tendency of a culture to separate into individuals with distinc-
tive characters.

The varieties of pneumococcus colonies which I am about
to describe show differences in morphology and in some of their
immunological reactions, which are similar to those found by
Arkwright in other bacterial species. They were discovered
in the course of my investigation of the mode of action of anti-
pneumococcus serum. On plating cultures of pneumococci
grown on homologous antisera, it was found that a number of
colonies which differed from the normal in appearance were
attenuated in virulence for mice.*

Attenuation was definitely associated with the production of
rough colonies, and these could be clearly distinguished from those
of smooth appearance which retained their virulence.

Culture Media.

Plate cultures. — Nutrient agar prepared with trypsinised meat
was used, with the addition of 5 per cent, chloroformed whipped
horse blood and 5 per cent, filtered horse serum. On this medium
isolated colonies of pneumococci grow up to 2 mm. in diameter,
and autolysis is often very slight.

The horse blood should be collected from a vessel directly into
a sterile bottle, since chloroform fails to get rid of gross contami-
nation. The blood cells, which rapidly deposit, are syphoned off
into medicine bottles, and about 2 per cent, chloroform is added :
it is useful to have some glass beads in the bottle to break up the
cells, which form a thick mass under the action of the chloroform.
The horse serum is filtered through a Berkefeld candle.

Blood broth. — Colonies from plates are grown in small tubes,
3 inches by \ inch, containing about 1 c.c. of equal parts of
whipped rabbit blood and trypsinised meat broth. Growth occurs
rapidly in the small bulk of medium, and can be recognised by
the medium becoming dark purplish in colour. The blood cells
keep almost intact in the ice chest for several weeks, and are
useful in showing the presence or absence of haemolysis due to
bacterial growth. After inoculation and incubation the cells
fall to the bottom of the tube, and a loopful of the deposit always
produces a good subculture of pneumococci in a tube containing
about 6 c.c. of ordinary broth.

Immune Serum. — The sera are prepared in rabbits from
virulent cultures grown for 4-6 hours in glucose broth. The
broth cultures are heated to 60° C. for an hour and centrifuged ;

* Laura Stryker (Joum. Exp. Med., xxiv, p. 49, 1916) had already
observed that repeated subculture of Types I. and II. pneumococci in
homologous serum caused marked reduction in virulence.

the supernatant broth is syphoned off, leaving about 15 c.c. out
of 100 c.c., in which the deposit is re -suspended. Rabbits receive
5 c.c. intravenously for each injection; the injections are. given
in series on 2 or 3 consecutive days and at weekly intervals, and
after about 2-3 months a usable serum is often produced.

For use as a culture medium the serum is distributed in small
tubes in quantities of 0*5 c.c, either neat or diluted as desired.
A few drops of a thick suspension of red cells (rabbit) are added
to favour growth and serve as an indicator.

Cultivation of Pneumococci in Immune Serum.

Several serological types of virulent pneumococci have been
grown in immune serum, though the majority of the observations
recorded have been made upon stock strains of Types I. and II.,
which were obtained from the Rockfeller Institute. The action of
the homologous serum upon a pneumococcus culture has been
controlled by simultaneous experiments with heterologous sera.
After each incubation at 37° C. overnight, the serum cultures
were sown on plates of blood agar so as to obtain discrete colonies.
With very few exceptions, which will be referred to later, no
alteration was observed in the characters of the colonies obtained
after growth in a heterologous serum, and smear preparations
showed that no agglutination had taken place, the pneumococci
being distributed uniformly throughout the medium. The
possibility that heterologous sera produced some biological
modification not associated with morphological change has not
been investigated. In the homologous serum growth took place
in the form of clumps and long chains or sometimes as translucent
gelatinous balls which deposited at the bottom of the tube. When
a first culture in serum was shaken and plated, in addition to the
normal S type of colony, a variable number of colonies appeared
which differed from them in size, appearance and consistency.
These were the R form. A second serum culture made from the
first yielded, when plated, a larger number of variants from the
normal type. After a third passage in serum the normal S form
was replaced by the R. Generally the pneumococci at this stage
had ceased to clump, and were tending to grow uniformly through-
out the medium. When a loopful of the first culture in serum,
which, as stated above, usually consists, after a night's incubation
of a mixture of R and S forms, is sown into plain blood broth,
serial cultures up to the 13th generation have remained a mixture,
the normal type in this instance showing a tendency to decrease
in numbers relative to the R type.

A single colony culture of either type, made at any stage of the
treatment with serum, as a rule, reproduces only the same type
when sown into plain blood broth. This rule is not without
exceptions, and occasionally, when the single colony culture of
the variant has been made after a single serum passage, a
reversion of some of the colonies to the normal smooth type
has been observed.

* 20179 B 3

The question of the stability of the R and S characters will
be considered later.

In order to produce the variant from a normal strain through
the influence of specific serum, it is necessary that growth should
occur. No alteration has been observed on plating a culture
that has remained in contact with the specific serum at ice-
chest temperature, or has been shaken with the serum at
room temperature. The variant, therefore, is a new generation
in the serum and not one of the original pneumococci altered
by contact with the serum. This point is of importance when
one compares the action of the serum upon the pneumococci in
the animal body and in the test tube.

Concentrated serum is not essential for the production of R
forms. A dilution of 1 in 256 has caused a partial change of a
smooth culture into a rough even with a single treatment. In
addition, the blood of a mouse, which has received a dose of
0-2 c.c. of serum, has been collected about 18 hours later, and
used as a culture medium. The pneumococci sown into it grew
as in a dilute agglutinating serum, and from one such culture,
after being incubated a night and then plated, several rough
colonies were produced.

Heterologous sera, as already stated, have generally no power
to produce rough colonies from a smooth culture and numerous
experiments have given negative results, even after several
passages. I have, however, found one exceptional instance. A
Type I. strain grown in a Type II. serum produced rough colonies
when plated: these remained rough after many generations in
blood broth. Until then it had seemed that the production of
rough colonies in a given serum might serve as a test of the
serological type. For example, the atypical II. A and II. B strains,
which are believed by the American observers to be related to
Type II., were grown in undiluted Type II. serum. Both strains
grew uniformly and not in chains, thus showing the absence of
agglutinins for them, and after five consecutive passages in
Type II. serum the colonies remained of the normal smooth
type. I have recorded elsewhere* evidence that the so-called
atypical Type II. strains are serologically distinct from Type II.

Description of the R and S Forms of Pneumococcus

Colonies.

All the serological varieties of pneumococci with the exception
of Type III. yield on the same medium colonies which are indis-
tinguishable. There is, however, only a difference in degree
between the Type III. colonies and the others, and sometimes the
colonies of Type II. and the atypical strains of Group IV. are
equally large and watery. The following description of the
normal type of colony applies only to strains which have been

* Reports on Pub. Health and Med. Subjects, Ministry of Health, No. 13,
1922.

kept in a virulent condition by repeated passage through animals
and maintenance in culture on favourable media. The medium
upon which the special differential features of the colonies are
best shown is the chocolate -coloured serum agar with chloro-
formed blood cells added. On ordinary agar, or on agar smeared
with fresh blood, the different varieties of colonies are much less
readily distinguished.

When the blood from a mouse which has died from pneumo-
coccus septicemia, or a young blood-broth culture, is plated
on the above medium and incubated overnight, all the discrete
colonies are of the same type. They are circular, dome-shaped,
with a smooth shiny surface and an almost watery consistency ;
when autolytic changes take place they become flattened, and
finally shaped like a button or draught. The characteristic
feature of the variant or rough type of colony is the surface,
which is dull or finely granular ; the colony is more coherent than
the smooth type, and may either break into pieces on being
touched or stick as a whole to the point of the needle. The
difference in appearance between the two varieties of colonies
on the same plate is often very striking, but it is essential that
the colonies should be well spaced. At first the rough variant
is generally smaller than the smooth colony : it is equally bile-
soluble.

Such are the two types of colonies which can be grown from
a normal virulent strain cultivated in homologous immune
serum. There is a close resemblance between the changes de-
scribed above and those described by Arkwright as occurring in
old cultures of various bacilli of intestinal origin. In the case
of pneumococci, as with these, rough colonies can be produced
also by growing cultures on agar media or in liquid media con-
taining glucose. Arkwright, as mentioned earlier, designated
his types of colonies S (smooth) and R (rough) respectively,
and this seems quite appropriate for the pneumococcal colonies,
the S being the normal and the R the variant. There is, however,
a slight distinction, which is probably one of degree, namely,
that the rough pneumococcal cultures grow uniformly in broth
and form stable suspensions, whereas Arkwright's R colonies
form clumps in any medium containing 0 • 85 per cent, salt solu-
tion. The rough cultures in broth are, however, a little more
granular than the smooth ; they clear more completely by centri-
fuging, and the deposit packs into a mass, which is not so readily
emulsified as the deposit from the smooth culture.

Stability of R and S Characters.

The smooth characters of the normal colonies remain unaltered
so long as the strains are kept virulent by passage through animals
and growth in blood broth. They persist even after prolonged
culture on solid media. For example, a number of cultures were
allowed to remain in sealed tubes of ordinary agar in the incubator
at a temperature of 37° C. At the end of three months all were

B 4

•till viable; on plating, some yielded apparently all smooth
colonies, and, where rough colonies were detected, they were few
in number. A rough colony culture, which has been produced
from a virulent strain by a single passage in immune serum, may
retain its rough characters for many generations in plain blood
broth and, as already mentioned, may subsequently revert to the
smooth type. This has occurred particularly with Type II., and
the reversion in cultural characters has been associated with the
recovery of virulence. Certain of the R colony strains, which
have been subcultivated for many generations in plain blood broth
and have been plated at intervals, have become in appearance
almost indistinguishable from the S strain, but they have
remained attenuated.

The change in the biological characters is of greater importance
than the morphological differences between the colonies. On the
other hand the characteristics of the R type obtained by three
passages through immune serum have been maintained for more
than 13 subsequent generations in plain blood broth. The activity
of the immune serum, i.e., the height of its titre, appears to have
an influence upon the permanence of the change.

Vibulence.

So far as my observations go at present the typical S colony
derived from a virulent culture is always pathogenic for mice
though one cannot affirm that all S colonies are equally virulent.
There is some evidence that a partial degree of attenuation is
associated with a corresponding tendency to the rough state, and
such partially rough colonies yield typically smooth colonies after
causing septicaemia in a mouse. The typical R colony shows, on
the other hand, a high degree of attenuation. When a virulent
pneumococcus is grown overnight in immune serum, a plate made
from the serum culture yields a mixture of R and S colonies. If
a colony of each type is selected and subcultivated for several
generations in plain blood broth, the S colony culture is found to
be as virulent as the original culture, e.g., killing mice in doses of
0-000,000,1 c.c. and 0*000,000,01 c.c. of broth culture, while the
R culture has no effect in doses of 0* 1 c.c. and 0* 2 c.c. (the latter
being the greatest amount inoculated).

Retested on mice after 18 generations in blood broth, the above
cultures gave similar results : the R culture failed to kill mice in
doses ranging up to 0*25 c.c. of blood broth culture, while the
S culture had remained apparently unaltered in virulence. If a
mouse dies of typical septicaemia after an inoculation of an appar-
ently typical R culture, the colonies grown from the blood are
invariably of the S type, that is, the culture has reverted. Occa-
sionally, however, a mouse may die after a large dose of R culture,
and the only organisms recoverable from the blood are a few
colonies of the R type. In such a case the peritoneal washings
do not give the characteristic agglutino-precipitin reaction with
the type serum prepared from the normal strain, and the blood of

the mouse produces no effect when inoculated into a second mouse.
The following example shows how repeated passage through im-
mune serum fixes the attenuated character. A colony of a Type
III. strain, virulent for mice in a dose of 0*000,000,1 c.c. of broth
culture, was cultivated for one night in Type III. serum ; a second
serum oulture was sown from the first. Both serum cultures were
plated when 24 hours old, and an R colony was subcultivated from
each in plain blood broth. (The R colonies of a Type III. strain
are readily distinguished from the large watery S colonies.) Each
culture was tested on mice. The R culture from the first serum
killed mice, and was found on being recovered from the blood to
have reverted to the S type. The colony culture from the second
serum was completely attenuated, and failed to kill mice in a dose
of 0-2 c.c.

Antigenic Characters of R and S Cultures.

My observations have not been numerous but they show that
the two types of colonies have different antigenic qualities. Sera
have been prepared in rabbits with the R and S cultures of both
Type I. and Type II. The results with Type I. sera, which I am
about to describe, were also obtained with the first samples of
Type II. sera. Later samples of the Type II. sera showed less
difference between the R and S sera, and on investigation it was
found that the R culture used for immunisation had partially
reverted to the S type. The Type I. sera were prepared with
strains which retained their special characters unaltered during
the period of immunisation. Three rabbits were immunised with
R culture, and one with S culture : the cultures were heated to
60° C, and were prepared freshly before each injection. All the
rabbits received identical treatment, 12 injections being given
intravenously during a period of five weeks. Samples of blood
were withdrawn eight days after the last inoculation.

Agglutination. — The serum from the rabbit injected with the
S culture agglutinated both S and R cultures up to a dilution of
1 in 160. The product of the reaction between the serum and the
S culture consisted of firm masses, which could not be broken
up by shaking. The R culture agglutinated with the formation
of loose clumps which were readily broken up on shaking to form
a finely granular suspension.

Of the three rabbits injected with R culture one only yielded
a good agglutinating serum. This agglutinated the homologous
R strain up to 1 in 640, and the clumps formed were of the same
loose character as those described above in the case of the reaction
between the R culture and the S serum. The S culture was
barely agglutinated in 1 in 10 with the R serum.

Precipitin. — The sera were tested upon broth culture which
had been freed from cocci by centrifuging and filtration through
a Berkefeld candle. When the S serum was mixed with the
filtrate of the S oulture, a precipitate was formed which aggre-

8

gated into a firm mass ; with the filtrate of the R culture there
was no visible reaction.

The serum prepared with the R culture gave a precipitate
neither with the R nor with the S filtrate.

Absorption of Agglutinin. — The serum of the rabbit immunised
with S culture lost after treatment with S culture all its agglu-
tinins, i.e., both for R and S suspensions. Treatment with the
R culture removed the agglutinin for the R and left unaltered the
agglutinating capacity of the serum for the homologous S strain.

The serum of the rabbit prepared with the R strain required
two successive treatments with the R culture before it lost the
whole of its homologous agglutinin. Similar treatment with the
S culture did not exhaust the homologous agglutinin from the
R serum, but reduced the titre for the R strain from 1 in 640 to
1 in 160.

Protection of Mice with R and S Sera. — The S serum, injected
intraperitoneally into mice in doses of 02 c.c, protected them
against a subsequent inoculation of doses of a Type I. strain
ranging up to 0*1 c.c. of broth culture, the virulence of which was
such that 0-000,000,01 c.c. of broth culture killed an unprotected
mouse within two days. Against the same strain of Type I. all
the R sera failed to show any protective power. The inoculation
of the three rabbits with R culture was continued in order to
ascertain whether more prolonged treatment would induce the
development of antibodies against the virulent strain. The
three rabbits were retested after they had been immunised for
a total period of four months, during which each rabbit received
26 intravenous injections. Their sera still failed to show the
presence of protective substances, when tested against a Type I.
culture which killed mice in a dose of 0*000,000,01 c.c. Two
other rabbits, which had been injected with the S culture for
five weeks and were tested at the same time, yielded sera which
protected mice against doses of 0-l c.c. of the above culture.

Influence on the Type of Colony by Growth
in R and S Sera.
A virulent Type I. culture was grown in R serum and in S
serum. There was no alteration in the character of the colonies
obtained on plates after three successive passages through the
R serum. On the other hand the culture grown in the S serum,
when plated after incubation overnight, yielded a mixture of R
and S colonies. One of each variety was subcultured in plain
blood broth, and was tested for virulence on mice. The S strain
killed mice in a dose of 0000,000,1 c.c. of broth culture, while
0*2 c.c. of the R strain proved innocuous.

Agglutinability of R and S Strains.

The clumps formed after the interaction between a virulent
pneumococcus culture and homologous immune serum are very
characteristic. On the addition of a 1 in 10 dilution of the serum

9

to pneumococcus broth culture there is an immediate turbidity, and
visible clumps quickly form, especially on gentle shaking, which
aggregate into a compact mass. An almost identical result, but
a less bulky precipitate, is obtained with a broth culture from
which the cocci have been removed by nitration. Thus, this
type of reaction, which is very specific for each serological type
of pneumococcus, depends on the formation of a precipitate
between the serum and a soluble substance in the broth. If,
however, the centrifuged cocci from a broth culture are resuspended
in salt solution, they form a compact mass with immune serum
similar to the above, although the soluble substance has been
removed with the broth. This appears to indicate that either
the external portion of the pneumococcus contains some of the
soluble substance in process of secretion, or both are of similar
constitution.

The attenuated R strain agglutinates with the same serum
in a different manner. If broth culture is used, agglutination is
slow, and fine granules form which settle to the bottom of the
tube. A more copious deposit is formed when a thicker salt
solution suspension of cocci is used. In both cases, however, the
clumps do not aggregate into a compact mass, but can be readily
shaken up into a finely granular suspension. The difference in
character between the reactions of the II and S strains seems
capable of a simple explanation. Since no precipitation or
turbidity occurs when the serum is added to a filtered broth
culture of the R strain, it is obvious that the R culture does not
form any soluble substance, and hence there is nothing to bind
the agglutinated cocci into a firm mass.

The R culture suspensions are rather less stable than the
smooth and, in addition, they are less specific in their agglutina-
bility, often showing some reaction with heterologous sera;
they continue, however, to react with the type serum to a higher
titre than with any other pneumococcal serum.

Mode of Action of Protective Antipneumococctts
Serum in the Animal Body.

The attenuation of virulent pneumococci by growth in
homologous immune serum in vitro is a phenomenon which is
probably related to the protective action of the serum in the
animal body and may help to explain it, though the two processes
are obviously not identical. In the test-tube the action of the
serum is shown in the altered virulence of the new generation of
pneumococci, hough the change is never complete in the first
passage through serum, the culture remaining a mixture of
virulent and non-virulent elements. Though pneumococci have
been recovered from an immunised animal at several stages
before their final disappearance it has not been possible actually
to demonstrate any change into rough colonies. In an immunised
animal the inoculated pneumococci have generally disappeared
from the peritoneal cavity within 24 hours; they remain alive

10

longer when inoculated into the subcutaneous tissues, where they
give rise to a localised abscess.

The following is the record of an unsuccessful attempt to
demonstrate the production of rough forms of pneumococci in an
immunised mouse. A virulent Type I. culture in broth was
mixed with an equal quantity of a Type I. serum and incubated for
half an hour, after which the mixture was inoculated intra-
peritoneally into two mice. A small amount of peritoneal fluid
was withdrawn from each mouse one and two hours later, part of
which was sown on plates. Only smooth colonies of virulent type
grew, but mice inoculated with the rest of the fluids remained well.
The two original mice were killed 20 hours after inoculation and
the peritoneal cavities were washed out with salt solution. The
centrifuged deposits from the washings were sown on blood agar
plates ; no growth of pneumococci was obtained.

If there is an intermediate stage in the mechanism of protec-
tion during which rough attenuated pneumococci, that is, cocci
which have not developed the part of their external structure
associated with the secretion of soluble substance, are produced
in a passively immune mouse, it is clear that they would readily be
destroyed, since even large doses of these R cocci inoculated into
an unprotected mouse quickly disappear.

On the other hand, it may not be necessary to postulate such
an intermediate stage. The immune serum may protect the
mouse by depriving the pneumococcus of its virulent potentialities,
possibly by precipitating the external covering associated with
capsule formation and the secretion of soluble substances. The
pneumococci so affected would become the prey of the phagocytes.
They would, if removed from the animal before destruction, be
capable of multiplying, and as the exposure to the immune serum
had been brief, they would produce normal virulent descendants.
But if within the animal body some sensitised pneumococci
escaped the attacks of the leucocytes and divided, the descendants
would remain under the influence of the serum, in which case the
property of the serum to produce progressive attenuation might
be brought into action.

Discussion of Results.

1. The first point of interest which emerges from this study
of S and R forms of pneumococci is the fact that a definite
morphological criterion has been found to distinguish between
virulent and attenuated strains. Further evidence of attenuation
is furnished by serological reactions, in particular by the loose
character of the clumps in the agglutination test, and by the
absence of specific soluble precipitable substance. Given these
data, it will be possible to decide whether a culture is in the
condition suitable for the production of disease.

2. An important practical corollary will be that, for the
production of good immune sera, strains should be used which
consist entirely of S forms. Immunisation of rabbits with the

11

R form has up to the present stage of the investigation failed
to stimulate the production of protective substances against a
virulent strain, whereas in the same time the injection of the S
form produced a strongly protective serum.

3. As regards the influences which cause attenuation of
virulence, I have found that this result can be produced invariably
by growth in specific immune sera. I have also noted that R
forms occasionally appear after prolonged residence on ordinary
solid media. It is possible that similar changes may arise under
natural conditions, as, for example, during convalescence from
pneumonia. The study of the various influences which cause
exaltation or attenuation of virulence may be facilitated by
observing the occurrence of S and R colonies under different
conditions.

4. This alteration from an S to an R type appears to be a
natural tendency with many species of bacteria. It can, I think,
be attributed to degenerative changes and is associated with the
loss of certain antigenic qualities. For example, while studying
the serological types of meningococci,* I subcultured separately
40 colonies of identical appearance and tested them against an
agglutinating serum prepared with the whole strain. With one
exception, all the daughter colonies absorbed the homologous
agglutinin. The exceptional colony, although it agglutinated as
well as any of the others, failed to lower the titre of the serum
for the parent strain. This last result I considered to be evidence
of a diminution of antigenic complexity. Now the relationship
of the R pneumococcus to the S is in some respects similar
to that of the variant meningococcus to the whole culture. A
serum prepared with the stock culture or with the S form of
pneumococcus agglutinates both R and S to the same titre.
The S form removes the whole of the agglutinin both for R
and S, but absorption with the R form has no effect on the titre
of the serum for S. The S and R possess related antigens,
as is shown by the agglutination test, but the R lacks an antigen
which is contained in the S. Before it can be said that the R
is simply an S form from which part of the latter's antigen
has been removed, it must be shown that the R type of antigen
is contained within the S form. A serum, prepared with R.
and agglutinating that form up to 1 in 640, barely agglutinates
the S form ; if present in the S form the R antigen therefore
is not available, or its union with the R antibody does not
result in agglutination of S. Still, repeated absorption of R
serum with S reduces somewhat the agglutinin for R, thus
indicating that there are receptors of the R type present in the S.

5. What are the distinguishing features of the part of the
antigen of the pneumococcus which is concerned in virulence ?
As one would naturally expect, in view of the intimate association
of a capsule with virulence, it appears to be the outer portion

* Journal of Hygiene, XVII., p. 166, 1918.

12

of the pneumococcus which is modified in the change from S
to R. After growth in immune serum the virulent pneumococcus
colony loses its smooth shiny surface and almost watery con-
sistency and becomes dull and finely granular. That is to say,
the individual pneumococci tend to adhere and are not separated
by some substance which, probably on account of its being
soluble in salt solution, permits the formation of a very fine
stable emulsion. In addition, the attenuated R form no longer
secretes in young broth cultures a soluble substance which reacts
with an antiserum from a virulent strain. There is some evidence,
therefore, that the R form, which is undoubtedly derived from
the S, is differentiated from the latter by the loss of that part
of its antigen which is concerned with virulence, capsule formation
and secretion of specific soluble substance. The R form may
bo considered to be a stage in the degeneration of a pneumococcus
from a virulent complex type to an attenuated form with a
simpler antigenic structure.

6. The next observation of interest is that this loss is not
necessarily permanent. Though in certain cases, as in old
cultures and after repeated serum treatment, the R forms are
apparently stable, it has been found, under other circumstances,
that this alteration in an important biological function can
apparently be reversed, not only by animal passage but also
by continued subculture in blood broth. After many generations
the R pneumococci assume again their smooth colony formation
and at the same time their virulence is restored. Can it be
said, in explanation of such results as these, that the original
R colony contained a few virulent pneumococci which escaped
notice in the earlier subcultures ? This hypothesis, in view of
the negative results of the periodical tests for virulence, cannot
be regarded as probable. Yet it is difficult to understand how
an attenuated organism can recover its virulence simply by
repeated subculture without animal passage. A somewhat
similar event has been described by Bail0 in his work on the
formation of capsules by B. anthracis. If a culture of anthrax
bacilli is exposed to a temperature of 42° C. for a period which
falls short of depriving it completely of the power to produce
capsules, certain strains are formed which produce a mixture
of colonies. Some of these colonies produce cultures all of which
invariably fail to form capsules, while others produce in addition
a few typical capsule-forming bacilli. This result, he thinks,
depends upon a deficient inheritance of the capsule-forming
substance, so that an individual bacillus is able to endow only
one of its descendants with a sufficient amount to produce a
typical capsule -forming strain.

7. Does this investigation throw any new fight upon the pro-
tective action of pneumococcus immune serum ? According to
Neufeld's bacteriotropic theory the pneumococci after inoculation
into an immunised animal become sensitised and are then des-

* Centraibl. f. Bakt., Orig., LXXIX., p. 42S, 1017.

13

troyed by phagocytosis. Experiments in vitro provide supporting
evidence, since virulent pneumococci are not ingested by leucocytes
except in the presence of immune serum, and the leucocytes are
thought to play an essential part in the mechanism and not to act
merely as scavengers.

Perhaps the above observations on the attenuating action of
immune serum may serve to amplify Neufeld's theory, the sugges-
tion being that the action of immune serum is not entirely bacterio-
tropic. A certain number of pneumococci may not be directly
subjected to phagocytosis in the presence of immune serum.
These may give rise to new generations in the animal body, but
such bacteria would, under the influence of the immune serum,
be subjected at the onset of growth to rapidly progressive attenua-
tion and would soon be destroyed.

Summary.

1. Virulent pneumococci become attenuated by growth in
homologous immune serum.

2. The loss of virulence is associated with an alteration of the
immunological characters.

The attenuated pneumococcus does not stimulate the pro-
duction of protective substances in the blood of inoculated
rabbits.

3. There is a morphological distinction between the colonies
of the virulent and the attenuated pneumococci. The virulent
colonies are smooth and the attenuated rough in character, the
distinction being similar to that between the R and S varieties of
B. dysenterise, &c. described by Arkwright.

4. The S pneumococcus produces in young broth cultures a
specific soluble substance ; the R pneumococcus does not.

5. The S form agglutinates with specific serum, producing a
firm gelatinous precipitate ; the R form produces loose clumps
which are readily shaken up.

6. An R strain may revert in all respects to the S type, or
may remain unchanged after many generations in subculture in
plain blood broth. On the other hand, the morphological dis-
tinction between the R and the S forms may tend to disappear,
while the immunological differences persist.

7. The attenuation of pneumococci by immune serum in the
test-tube indicates that the mode of action of the serum in a
protection experiment is a direct one upon the pneumococcus.

A double action may be suggested : —

(a) The serum may disorganise the biological functions
of the pneumococcus by precipitating the capsule, thus
inhibiting the secretion of anti-leucocytic substances and
rendering it temporarily harmless.

(b) When pneumococci divide in the animal body in the
presence of immune serum, it is suggested that the influence
of the serum may cause progressive attenuation of subse-
quent generations.

November, 1922.

14

a— BACTERIAL VARIATION AND TRANSMISSIBLE

AUTOLYSIS ; THE RELATION OF BACTERIAL ENZYMES

TO BACTERIAL STRUCTURE.

By Arthur Eastwood, M.D.

Introduction.

Careful investigation of cultures, grown on ordinary or on
selective media, has shown that variations are of common occur-
rence in most bacterial species. The variants, whilst retaining
the general characters of their species, exhibit differences in
one or more of the following attributes* : — appearance of the
colonies, morphology of individual bacteria, motility, fermenta-
tive capacities, capacity for growth, antigenic properties, and
virulence.

Of the principles which determine these changes very little
is known, but certain laboratory data seem to be clearly established.
In the first place, variants can be produced in cultures derived
from a single cell or from single colonies, a procedure which has
been repeated so often that there can be no doubt as to its
accuracy; thus one has no difficulty in answering the question,
Have you really produced a variant, or was the " variant "
present in small numbers, which were unrecognised, in the material
with which you started ? The answer is that variants undoubtedly
arise de novo and that this fact has been proved by properly
controlled experiments, which show that variants are definitely
attributable to some change associated with bacterial growth.
In the next place, though variants observed in the laboratory
are not all of equal importance and though some of them may be
merely artefacts, in the sense that they are attributable to causes
not likely to be operative in nature, the main fact remains that
many of these variants are of great interest, because they throw
light on the biological properties of bacteria and are thus related
to the problems of infection, resistance and immunity.

To take the most obvious instance. A strain is split up
into two types, the " normal " and a variant ; though each grows
equally well in subculture, when tested on animals it may be
found that the former has retained its virulence whilst the latter
has lost it. This observation is clearly of interest in discussing
the factors on which virulence depends. Then one may go a
step further. Under the operation of certain influences in vitro,
a culture may be separated into two types, the " sensitive "
and the " resistant," which do not grow equally well, even in the
test-tube. The former is much less viable and the sensitive

* I am assuming, provisionally and for the sake of simplicity, that the
attributes of the normal bacterium are clearly and accurately defined.
This, however, is not always the case ; it is sometimes difficult to say which
type is " normal " and which is " a variant."

15

individuals which are just viable tend to subdivide into non-
viable elements and may, eventually, be eliminated. This
artificial method of producing, and finally eliminating, a " sensi-
tive " variant, under the influence of what has been called a
" lytic principle," calls for careful consideration, because it may
have its counterpart in the animal body, as one of the many and
highly complex factors in the mechanism of resistance to bacterial
invasion.

As bacteriologists are familiar with the more important work
on bacterial variation and transmissible autolysis, there is no
need for me to reproduce a detailed review of the literature.

For the immediate purpose of this report a brief historical
note will suffice. Opinion on the significance of bacterial variants
seems to have been passing through the following phases : —

(1) In ordinary routine work, when an organism is
plated out for the purpose of isolating and propagating
a pure culture, the bacteriologist selects what he knows from
experience to be a typical colony and does not trouble about
other colonies which look more or less different from his
idea of the normal. The question of variants he regards
as negligible.

(2) But patient comparative study of different colonies
derived from a single strain, which is known to be pure, has
shown that, with most bacterial species, the occurrence of
so-called " aberrant " forms is not exceptional but is the
general rule and that, therefore, the term " aberrant "
may not be justifiable. This is disconcerting, because it
means that the forms with which the bacteriologist actually
has to deal are not " ear-marked " by nature as " normal "
and " variant " but are simply " varieties," each of which
may be equally normal. Consequently, one's ideas of what
is " typical " and what is a " variant " need justification
and may require revision. At this stage it is realised that
the question of variants is not negligible ; it has to be taken
seriously, because much confusion arises from failure to
recognise actual, or potential, differences in individuals
amongst the bacteria comprised in an ordinary " pure
culture."

(3) The study of variants was diverted into a new channel
by Twort, who discovered what is now known as " trans-
missible lytic substance." He first found this substance
in growths of micrococci cultured from vaccine virus; it
consisted of glassy, transparent material which lysed the
micrococci and was transmissible from culture to culture.
Subsequently d'Herelle discovered similar material, which
he obtained, in the first instance, from the intestinal excreta
of a dysentery patient. These observations have aroused
much controversy as to the nature of the lytic substance.
D'Herelle, for example, has always maintained that it is a
living, ultra-microscopic virus and calls it a " bacteriophage."

16

Bordet thinks that it is an autolytic enzyme produced by
the bacteria which undergo lysis and that it is due to " a
nutritive vitiation primarily started by external influences,
an example of which may be the contact with a leucocytic
agent." Bail has suggested that it consists of minute
particles of the bacteria themselves and that these particles
will live and multiply, not independently but when provided
with nutrient material from living bacteria. Otto and
Winkler regard it not as a living virus but as composed
of minute fragments of bacterial protein, arising from the
disintegration of living bacteria and endowed with the
properties of enzymes. This stage is characterised by
controversies between upholders of conflicting theories and,
generally, by the assumption that the " Twort-d'Herelle
phenomenon " is an entirely new discovery and differs
from previously observed factors concerned with bacterial
variation.

(4) Since then a reaction has set in. More attention
has been paid to the fact that " lytic substance " is not
merely an agent which destroys bacteria ; it is also a means
of producing bacterial variants. Moreover, it has often
been found that some of these variants closely resemble
variants described long ago, before " lytic principle " was
discovered. Further, it is suggested that "lytic principle "
is not a new discovery but simply a revival — supported by
new experiments — of old ideas and observations about the
lytic properties of bacterial enzymes ; and the balance of
opinion seems to be definitely against d'Herelle's view that
it is attributable to a living virus which acts as a
" bacteriophage."

For my own part, I agree to a large extent with phase (4).
In particular, I think that " lytic phenomena " are closely
associated with other facts about bacterial variation where lysis
may not be demonstrable, and that the former should be considered
in conjunction with the latter as part of the same general problem.
But, whilst it is desirable to link up new work with old, wherever
it is possible to do so, I would not go so far as to suggest that
" lytic substance," which acts only on living and actually growing
bacteria, may be identified with those previously known bacterial
enzymes which are much less restricted in their sphere of action.

In concluding this note, I may give a few references for
readers who desire a fuller historical account. Penfold (Journ.
Hyg., xi-xiv, 1911-14) has written extensively on bacterial
variants and has suggested changes in the enzyme chemistry of
the bacterial cell as an explanation of his observations. Arkwright's
article (Journ. Path. & Bad., xxiv, p. 36, 1921) is important;
it describes his observations on " rough " and " smooth "
colonies and brings them into line with other investigators' work
on mutation. The question of the " lytic principle " was discussed
at the last meeting of the British Medical Association and is

17

reported, under the title " The Bacteriophage (Bacteriolysin),"
in the British Medical Journal for August 19th, 1922. More
recently (November, 1922), Medical Science Abstracts and Reviews
has summarised the literature on this subject under the title
of " The d'Herelle Phenomenon."

I think it will be agreed that the whole subject is at present
in a state of much uncertainty. There is a multitude of observa-
tions on bacterial variants, but no general explanation of the way
in which they arise ; and it is often impossible to find any cor-
relation between a particular type of variant and a particular
kind of disturbing influence. Work on the " lytic principle " is
particularly difficult to follow, because there are so many con-
flicting theories as to the nature of this principle. As the whole
subject is very far from being thoroughly worked out, this
confusion is only what was to be expected at the present stage.

In the hope that further discussion may help to clear the
ground, I put forward the following hypothesis (pp. 17-23) as a
possible explanation of some of the phenomena associated with
bacterial variation. It seems to me that the essential problem
concerns the physiology of the bacterial protoplasm. After
developing my hypothesis on this basis, I proceed to illustrate it
by recent laboratory data.

Facts to be Correlated.
The main facts to be correlated are the following : —

(1) Bacteria "breed true" to their species; any varia-
tions which occur are within the limits of the species.
There is practically no evidence which would make it
worth one's while to give serious consideration to the
possibility of transmutation of species.* It need not be
assumed that, when the living bacterial cell divides into
two, each portion necessarily retains the species characters ;
but it does appear to be a tenable hypothesis that any new
individual which does not retain characters requisite for
identification of its species is not viable, the limits of
variation being thus determined by the limits of viability.

(2) The variations which are found are not, as a rule,
irregular or haphazard but tend to some one definite
direction. This is particularly the case when the change
is«due to a single influence and not to a variety of factors.
The first stage in the process is generally the production
of a mixture. The " normal " strain (a) produces a mixture
of (a) forms, which all resemble the original, and of (b) forms,
which are alike among themselves and possess one or more
attributes differentiating them from (a). In imagining
how this result comes about, there are two possibilities to
consider. An (a) form may subdrv ide into an (a) and a (6) ;

* I do not propose to enter here into academical questions about the
precise definition of " species " as applied to bacteria.

18

or it may subdivide into two portions, only one of which,
either (a) or (6), is viable. I think this latter possibility is
worth bearing in mind. Continuation of the influence
which produced the mixture may result in elimination of
the one form and survival of a pure culture of the other.

(3) Starting with a pure strain of b as the normal,"
it may be possible to grow it under conditions which cause
it to throw off variants (c), all resembling each other and
differing in the same respects from 6. And this may not
be the final limit to the possible number of variants.
By appropriate treatment of c, another variant may be
split off, and so on. But a multiplicity of variants does
not mean that they are produced in a haphazard sort of
way ; they seem rather to emerge one at a time, as a series
of consecutive reactions between the modified bacterium
and its environment.

(4) Under special conditions .the characters of the
variant recovered from b may be those of a, i.e., reversion
to the original type may occur. Such reversion is most
commonly observed when the original influence which
converted a into b was in operation only for a short time.

(5) In the initiation of bacterial variation, a bacterium
(A) may divide in such a way that one half (the variant)
is not viable, whilst a fellow bacterium (B) may divide
into halves which are each non-viable. If all the bacteria
behaved like B, the production of variants would mean
death of the culture. But suppose that, whilst a large
majority of the bacteria behave like B, the remaining
few behave like A; the result will be that disappearance
of the variant is associated with scanty growth of the
more resistant survivors of the normal individuals. This
appears to be what happens in the production of " autolysis "
associated with the growth of bacteria. It seems to be a
special example of bacterial variation, due to an influence
which is not an ordinary lysin, because it acts only on
bacteria in active growth; it is a selective stimulus, the
nature of whioh will be discussed later, causing the bacteria
to subdivide into non-viable elements with the survival
of some resistant individuals. For the sake of simplicity,
I have stated the case as above; but there is usually an
intermediate stage before the variants disappear entirely;
some of them are just viable, though tending to subdivide
into non-viable forms; if removed from the lytic stimulus
at this stage, they may be cultured as a variant charac-
terised by its high sensitiveness to lytic influence.

Correlation of the Facts.

Is there any interrelationship between the data set out above ?
Perhaps a clue may be found in considering the intimate way
in which two essential processes of bacterial life are dependent

19

on each other, viz., the enzyme action which breaks up the food
supply into particles suitable for assimilation and the synthesis
of these particles to form living protoplasm.

Considered apart from their association with living matter,
enzymes are catalytic agents and their action involves an inter-
mediate stage, during which they form an unstable combination
with the material (substrate) undergoing digestion ; then follows
dissociation of the digested substrate from the enzymes, leaving
the latter free to attack fresh substrate. But this is not neces-
sarily the whole story. Conditions may supervene which tend
to stabilise (instead of dissociating) the union between enzyme
and digested food material, the result being that enzyme action
ceases to be progressive and is replaced by a synthesis between
substances formerly acting as enzymes and material upon which
they acted.

Conditions of this latter nature seem to be of particular
importance for cellular growth, in explanation of which it is
necessary to postulate a balance of two opposite factors —
(1) enzyme action, and (2) inhibition of this action; without the
aid of (1) the cell would die from starvation ; without the
restraining action of (2) it would die from " autolysis." Applying
this conception to the enzyme action of bacterial protoplasm,
it would appear that bacterial substances which act as catalytic
agents are also concerned with the synthesis of bacterial protein,
and that bacterial protoplasm may be regarded as a complex of
enzymes and the products of enzyme action, a complex which
involves the synthesis of these products {e.g., amino-acids) into
proteins.

Starting with this idea that the cellular enzymes of bacteria
are not merely catalytic agents acting on a substrate, but are
also the nucleus upon which bacterial protoplasm is built up,
it would seem that, when a bacterium begins to grow, catalytic
action upon its environment is at first predominant; then it
passes through a critical phase which fixes or stabilises the
constituents of bacterial individuality, a phase which is marked
by the transition from catalytic action on the raw substrate
to synthetic action on the elaborated product. When full
development is followed by division, these processes are started
afresh.

As bacteria breed true to species, it must be assumed that,
in the mechanism of growth and reproduction, there is a constant
repetition of the same processes, with identity of catalytic agents,
of products ready for assimilation, and of their synthetic combi-
nation. This precise uniformity becomes easier to understand
with the help of the idea that both catalytic and synthetic
action are attributes of the same protoplasmic substances.

Then where is one to find opportunities for variation (within
the limits of species characteristics) ? Here the incidence of
what I have termed the " critical phase " in the transition from
catalysis to synthesis may be worth considering, as providing

20

such opportunities. When a young culture is growing under
favourable conditions, without the production of any variants,
it may be presumed that the stabilising influence which promotes
synthesis acts at exactly the right time and in the right way.
But under less favourable circumstances, this influence may
begin to operate (a) a little too soon, or (6) a little too late,
or (c) much too late ; and the consequences may be, respectively
(a) failure of some of the bacteria to develop their full biological
properties, e.g., loss of some function or of some antigenic con-
stituent, (6) the acquisition (after full development of biological
potentialities) of some additional chemical groups which tempo-
rarily mask one or more of these potentialities, e.g., disappearance
and subsequent return of some biological property, (c) failure
to build up the new protoplasm requisite for growth, with
" autolysis " as the result. And one might amplify these possible
causes of variation by imagining that the stabilising influence
did not act " quite in the right way," i.e., that it caused some
deviation from the normal molecular or colloidal arrangements
for the synthesis of chemical groups into protein.

Is it possible to explain in more definite terms the idea that
certain variations in bacterial life are attributable to irregularities
in the action of a " stabilising influence " ? I think it is; but
one must not be expected to perform impossibilities. No
physiologist can explain all the activities of living protoplasm;
and bacteriologists are equally unable to give the precise reasons
why a bacterium grows or why it grows in a particular way.
All that can be done is to note certain influences which probably
participate in the phenomena of bacterial life.

I propose first to describe some of these suggested influences
in general terms and then to illustrate them by concrete examples.*

It is convenient to consider separately — I. The initial stimulus
to variation and II. The propagation of the variant.

I. The causes which may initiate variation are many and
diverse. Roughly and provisionally they may be divided into
(a) those which are obviously non-specific, i.e., where there
can be no structural relationship of the " lock and key " type
between the stimulus and the bacterial protoplasm, and (b) specific
influences, where the stimulus has, directly or indirectly, a special
selective action, due to structural relationship, upon the bacterium
or upon the environment of the bacterium.

Under (a) are included a large number of purely physical
influences, such as a slight change in temperature, or in the
reaction of the medium, or in the colloidal balance between the
bacterial constituents and their environment. Exactly how
each influence operates cannot be explained, but it is known that
the normal arrangement of the particles which are being synthe-

* Perhaps readers not already familiar with the subject will find the
argument easier to follow if they take the concrete examples (pp. 23-33)
first. This, of course, is not the correct logical sequence, as the examples
are not the basis of the more general statement, but merely illustrations
of it.

21

sised by the bacterial protoplasm is often disturbed by purely
physical causes.

A good example of (6) is the production of a variant by growth
in homologous immune serum. Here the conditions are simpler
and it is possible to explain the stimulus which initiates the
variation as being directly due to the specific antibodies in the
serum. In this case it is reasonable to postulate that, when
digested food particles are being synthesised into protoplasm,
the antibodies find in some of those particles their appropriate
antigen and " pick them off " ; the new bacterium is synthesised,
but it is an impoverished bacterium (a variant), because it has
been robbed of some of its antigenic components. To take
another example, the products of bacterial growth are sometimes
highly specific for the bacterium in question ; when these products
are removed by filtration, growth of the bacterium in this medium
(the filtrate) may initiate the production of a variant. How this
comes about is not really known; one can only guess. Perhaps
it may be said that particles in the bacterial filtrate have receptors
identical with those of the living bacterial protoplasm and there-
with " pick up " particles ready for assimilation by bacteria
in the nascent stage of growth ; such bacteria are impoverished,
i.e., they are a variant; they may be so much impoverished
that their offspring are non- viable.

I have said that the initial stimulus may be (a) non-specific,
or (6) specific. But these are not the only alternatives. It may
be partly the one and partly the other ; or it may act like (6) on
one occasion and like (a) on another. To revert to the last
example, a bacterial filtrate may be so highly specific that it
can only produce a variant when acting on its homologous strain ;
other filtrates may act on all strains of the same species, others
on allied species, others, again, on some species which have no
relationship to the bacterium producing the active principle
contained in the filtrate.

II. When once a new variant is formed, its propagation is
obviously due to a spec 'fie influence, which is the special attribute
of the modified bacterial protoplasm. The new arrangement
of particles and side-chains usually tends to repeat itself auto-
matically, from generation to generation and from culture to
sub-culture, without the need of any external influence. But
in the type of variation which is associated with " transmissible
autolysis " an external influence is still in operation. Two pro-
cesses are at work simultaneously ; there is renewal of an external
stimulus (now specific) for the initiation of the variant, and there
is the tendency for the variant to multiply automatically, though
with the production of many non- viable forms. To put it briefly,
the starting point is a stimulus, either non-specific or specific,
causing the formation of variants, some of which are non- viable.
These latter disintegrate into particles which act as specific stimuli
(like the bacterial filtrates mentioned above) and change more
normal bacteria into variants ; many of these are non-viable and

22

disintegrate, with the liberation of wore "lytic substance";
so the process is repeated, the net result being rapid " growth "
of "lytic substance." At the same time those of the variants
which are viable subdivide into a varying proportion of viable
and non- viable forms, from the latter of which the supply of
" lytic substance " is again increased.

Experiments with lytic substances raise interesting, and often
puzzling, questions as to the interplay between specific and non-
specific factors in the production of variants.

Here are some instances. A lytic substance derived from
a particular strain may (1) act on all strains of the same species,
or (2) it may fail to act on some of the strains. Is (1) due to
identity, and (2) to lack of identity in the receptor apparatus ?
That would seem the simplest explanation. Further, when
the lytic substance acts on some other species, as well as on the
species from which it was derived, is that due to community of
" group " receptors ? This may be the case, but it is not so
easily acceptable. Here " group " receptors, in the purely
chemical sense of the term, may have to give way to more physical
conceptions; for example, the colloidal condition of bacterial
protoplasm may differ in different bacterial species and may be
influenced in some cases, but not in others, by the finely-divided
colloids in the filtrate containing a particular lytic principle.

Still, it is worth considering how far it may be possible to
apply the " receptor theory." Different stimuli may be equally
effective in causing a culture to produce lytic substance. For
example, a stimulus, a, may cause a normal culture of bacillus C
to produce lytic substance c1, which may be transmitted to a
fresh normal culture, and so on, indefinitely. Another stimulus,
6, may produce the same effect on bacillus C, yielding a supply
of lytic substance c2, which may be propagated ad infinitum.
One might suppose that, after several transmissions through
identical cultures of normal bacillus C, c1 and c2 would be com-
pletely identical with each other. But this may not be the case.
If stimulus a was a filtrate from a culture of bacillus A (an organism
different from C), whilst stimulus b was derived from some other
source, and if the lytic properties of the propagated c1 and c2
be tested on a culture of A, it may be found that c1 produces
lysis, and that c2 fails to do so. How does c1 " remember "
its affinity for A ? Did a cause the protoplasm of C to break
up in a special way, viz., into particles containing receptors
for both C and A ? And was this principle perpetuated in the
transmission of c1 from culture to culture of C ?

This idea of linking up of two (or more) kinds of receptors
leads to a further question. By immunisation with bacterial
filtrates containing a lytic principle, " antilytic " sera, i.e., sera
which neutralise or inhibit lytic action, may be produced. Some
of these neutralisation experiments suggest that one may be
dealing with a combination of receptors (A and B), of which one
(A) is dominant and the other (B) is in abeyance or " masked."

23

For example, the lytic action of an active principle containing
both A and B may be neutralised by an anti-A serum but not
by an anti-B serum. Is this a valid explanation ?

Whilst it must be freely admitted that the facts about " trans-
missible autolysis " are very difficult to understand, it seems to
me that the most promising line of explanation lies in a develop-
ment of the suggestion that these data are merely special instances
of a general principle determining bacterial variation.

This view is in harmony with experiments which show that
the lytic principle is not merely an agent of destruction; it is
also a means of increasing bacterial resistance. In the exercise
of this latter property its action is progressive. First it divides
the bacteria into some relatively resistant forms (A), which
survive, and into more sensitive forms (B), a small proportion of
which also survive. The process is repeated with the descendants
of A and B ; the Weaker individuals die out and the stronger
survive, perhaps including amongst the latter a few more hardy
descendants of what were originally B forms. Carried to its
conclusion, the process results in the production of a pure strain
of highly resistant A forms, the variant (B) having died out.
Thereupon the pure A strain behaves like a normal culture,
though more robust than the normal because the tendency to
variation has been eliminated ; it is no longer lysogenic, because
the sensitive variants which, on disintegration, yielded the lytic
substance, have disappeared.

At this point it is useful to note that, by using the appropriate
laboratory method, one of two opposite results may be achieved
in the experimental modification of bacteria. Taking advantage
of the fact that a normal culture may be split up into resistant
and less resistant forms, it is possible to eliminate either of the
two and to produce at will, a pure (or relatively pure) form of
either the one or the other.

The practical importance of these laboratory results is that
they provide information about some of the factors which deter-
mine susceptibility and resistance to bacterial invasion.

They at once raise the question, Do similar influences deter-
mine the production of bacterial variants in the animal body ?

Significance of the Experimental Data. .

Examples of bacterial variation are very numerous and
present a wide range of differences. All that can be said about
many of them is that they are carefully recorded facts but their
significance is obscure and the causes which produce them are
unknown. It will be more useful, for the purpose of this report,
to select some examples of variants which are clearly of importr
ance in the study of infection and immunity, and to confine
one's attention to these.

Spontaneous Variations.

In the first place, valuable information has been obtained
from observations on variants which occur " spontaneously,"

24

i.e., without any effort, on the part of the bacteriologist, to produce
them.

Andrewes,* working with the four closely related organisms,
B. paratyphosus B and C, B. ceftrycke and the Neivport bacillus,
found that in ordinary young cultures of undoubted purity
there existed side by side two varieties which could be separated
by plating out. They differed antigenically in a very pronounced
manner. " The one group reacted well with the monospecific
serum, .... while they failed to agglutinate at all with
the allied group serum. The other group behaved in the
reverse manner, agglutinating well with the group serum,
though not to its full titre, while they reacted only weakly
with the monospecific serum." These differences in agglutin-
ability were confirmed by agglutinogenic tests. A serum pre-
pared with a " specific " variety was highly selective for specific
strains ; whereas a serum prepared with a " non-specific " variety
reacted with the homologous strain and with other non-specific
strains but " it fails to react with specific strains except the one
of its own type, and even here the titre is very low." The differ-
ences in agglutinogenic action were the main fact ; the slight
degree of overlapping was to be explained " on the supposition
" that specific strains contain a small amount of group antigen,
" and unspecific strains a small amount of specific antigen." It
is of particular interest to note that the only differences found
were those which were demonstrable by serological tests. " The
" phenomenon had nothing to do with the ' rough ' and ' smooth '
variants described by Arkwright (1921); all the colonies
' were in every other respect alike, and all gave uniformly
" turbid broth cultures." Furthermore, Andrewes found that
the same phenonenon might occur in the human body. " I had
" the opportunity of examining the primary spleen culture
" from a woman who had died of iErtrycke infection ; ten colonies
" were picked from this plate and equally fell into the two
" sharply defined groups." Two other facts are of importance.
(1) No intermediate forms occurred; the varieties always fell
into one or other of two distinctive groups, the " specific " or
the " non-specific." (2) On sub-culture, each type readily
changed into the other. " I have at times succeeded in obtaining
' two consecutive broth cultures in which the character was
" maintained pure, but far more commonly even the second
" broth culture shows a mixture of types, and I have only rarely
obtained pure characters on solid media."
The above results are clean-cut and are derived from experi-
ments which were free from complicating factors. They are
consistent with the observations of other workers who have shown,
by the use of the absorption method, that individual bacteria in a
pure culture may differ from each other antigenically, some being
less fully equipped than others in the attributes of complete
specificity. For example, F. Griffith states in the preceding

* Journ. Path, and Bact., XXV., p. 505, 1922.

25

report (p. 11) that, when he was working with meningococci,
he plated out a culture and examined 40 colonies which were
all identical in appearance, but one differed from the rest in failing
to lower the titre of the serum for the parent strain, i.e., it showed
•' a diminution of antigenic complexity." Schiitze,* again,
found that he could get " substrains." i.e., strains in which it
was shown by absorption that the " antigen mosaic " was
" defective," by simply making single cell cultures from a normal
strain. In both these examples, as in Andrewes' work, the
variants were obtained from young cultures and the only differ-
ences between the variant and the normal were those which
could be demonstrated by serological tests.

Andrewes' data are a particularly good example of what I con-
sider to be the earliest stage of spontaneous variation. It is found
in young cultures and consists of only a slight irregularity in the
synthesis of bacterial protoplasm, characterised by one feature
only — irregularity in the disappearance and reappearance of a
particular antigenic constituent.

I propose now to call attention to a more advanced stage of
spontaneous variation. Examples are to be found in the work of
Arkwright, Benians and Mary Cowan, to which F. Griffith has
made reference above (p. 1). In the present connection the
following points are worth noting. The production of the changes
noted by Arkwright seems to require a more prolonged influence
than is the case with Andrewes' variants ; old cultures have been
found especially useful for plating out colonies exhibiting the
characteristic modifications. And the changes in the variants
are more profound and more stable. Their colonies are distin-
guished by their " rough " or " smooth " appearance. The
source of the former, the " R " variants, " has been almost

invariably old broth or agar cultures. They may be obtained
" from most strains if a culture which has been kept at room tem-
" perature or in the incubator for a month or more is plated out."
And, though he found that " R " forms might sometimes be made
to revert to " S " by daily subculture, the two types were much
less liable to reversion than Andrewes' varieties. Like the latter,
the " S " and " R " forms " differ very decidedly in their agglu-
" tinating, antigenic and absorbing properties with specific sera,"
but a new feature is found in association with the " R " form ;

it agglutinates in 0*85 per cent, solution of sodium chloride
and in broth cultures it forms a deposit leaving the liquid clear."
With regard to his " S " and " R " forms of a Shiga strain, Ark-
wright stated that both " were fatal to rabbits when injected
" subcutaneously or intravenously in very small doses, but
" no attempt was made to find the M.L.D."

Benians' somewhat earlier observations are worth bringing
into line with Arkwright's, as the results are more or less similar.
He inoculated a guinea-pig subcutaneously with a mixture of
a typical (a) Shiga bacillus and mucilage of tragacanth; the

* Journ. Hyg., XX., p. 333., 1921.

26

animal's local lesion then acted as a culture tube in which the
bacillus was allowed to age. On plating after two months, he
obtained a mixture of (a) colonies and morphologically different
(b) colonies. When (6) was subcultured, it often threw off some
(a) colonies, " especially soon after it was first isolated." The
variant (6) " usually sedimented both in broth and saline " ;
it was inagglutinable and not agglutinogenic, but conferred active
immunity against both itself and (a) and was quite as pathogenic
as (a).

The special point of interest in Mary Cowan's subdivision of
streptococci into " rough " and " smooth " strains is that the
former were much less pathogenic than the latter.

To summarise the above, when the action of the spontaneous
modifying influence is prolonged, the following changes make their
appearance. The colonies formed by the variants are distinctive ;
there is more profound alteration in agglutinability and agglu-
tinogenic capacity; there is less tendency of the variants to
revert to the parent form, and, finally, there is loss of virulence.

In a still more advanced stage of spontaneous modification
there is the production of lytic substance.

Here are some examples. Bail* stated that on three occasions

he had obtained lytic substance directly from three old cultures

of Flexner's bacillus. Gildemeisterf declared that the presence

of lytic agent could be demonstrated in a special type of colonies

which he had described previously (1917) under the name of

" inconstant forms " (Flatterformen). He gave them this name

on account of their irregularity and their liablity to change in

subculture. He was working with intestinal organisms belonging

to the coli, dysentery, typhoid and paratyphoid groups and had

isolated his cultures from faecal material. Otto and MunterJ

also noted that the lytic principle could be obtained from old

cultures (aged 3 weeks to 3 months) without the aid of the animal

body. Gratia, § again, has obtained similar results by merely

allowing a normal culture of B. coli to grow old. ' ' An old agar slant

" of Bacillus coli shows an uniformly dull film on which appear

" very distinctly, here and there, small vitreous colonies." These

colonies produced a growth " which possesses a great resistance

to the lytic agent," whereas the " dull film " was found to be dead.

" At the same time we have had the good fortune to isolate

from a subculture of the original strain a colony of organisms

which are, on the other hand, extremely sensitive to the lytic

agent." The two types grew differently in broth, and the

sensitive type was also distinguished from the other by being

non-motile and less virulent. Ledingham.|| after remarking that

some strains of bacteria " show lytic changes normally," says : —

I attach great importance to this fact. Dr. Lepper, at the

* Wien. klin. Wochenschr., p. 447. 15th Sept., 1921.
t Berl. klin. Wochenschr., p. 1355. 14th Nov., 1921.
j Deutsch. med. Wochenschr., p. 1579. 29th Dec, 1921.
§ Journ. Exper. Med., p. 115, 1921.
|| British Med. Journ., p. 297. 19th Aug., 1922.

27

" Lister Institute, recovered from the urine of a case of pyelitis a
" coliform strain which behaves in culture as if it were a mixture
" of bacilli and ' bacteriophage.' "

I do not think there is any need to quote further instances.
If spontaneous production of lytic substance directly by the
bacteria had only been observed in rare instances, the true explana-
tion might have been that this substance was not produced de
novo but was present, though unnoticed, along with the strain
used as the starting point of the observations. But this objec-
tion will not hold, because these results have been observed
repeatedly by different investigators, who have started with pure
cultures showing no evidence whatever of the action of a " lytic
principle " when growth was allowed to take place under normal
conditions.

Experiments with Agents capable of producing
Variants.

I take as my starting point the production of pneumococcal
variants by growth in immune serum, which is described by
F. Griffith in the preceding report. Here the influence which
produces the variant can be identified at once; it is the action
of the specific " type " antibodies on the growth of the homologous
strain. The variant is viable in vitro but not in vivo ; and its
incapacity to thrive in the animal body is shown to be due to loss
of those elements in bacterial structure which are associated
with capsule formation and production of " specific soluble
substance." Hence the action of the serum in protection tests
is readily explained; the serum produces a variant which, owing
to these defects, is not viable in the animal body. Possibly
the therapeutic action which has been demonstrated for Type I
antipneumococcal serum is also to be explained, or partially
explained, on the same principle.

At the same time it must be recognised that this principle is
only applicable under special circumstances and does not go
very far as a general explanation -of the more important facts
about immunity reactions and the production of variants. For
example, it is impossible to maintain that recovery from pneu-
monia, without serum treatment, is due to manufacture, by the
patient, of those antibodies which are demonstrable in artificially
prepared immune sera. And similar difficulties arise in attempting
to show that resistance to infections with other species of parasitic
bacteria is due to demonstrable antibodies.

Hence, if the results cannot be explained as due to known and
demonstrable antibodies, one is led to consider whether there
may not be some other specific factor which operates in the living
body and converts the invading organism into a non-viable
variant. Antibodies, it should be remembered, are not the only
influences which are specific for bacteria. Bacterial attributes,
though they may be antigenic {i.e., capable, as foreign protein, of
giving rise to antibodies) may also act selectively, and even

28

specifically, in other ways. They may produce non- viable variants
by direct action upon bacterial growth, not by the indirect method
of stimulating the animal body to produce antibodies which
interfere with that growth. Experimentally there are many ways
of producing such variants without the employment of antibodies.
One of these methods, the use of " lytic substance," is of particular
interest, because what is seen to take place in the test-tube may
have some counterpart in the reactions going on in the animal
body.

The first fact of importance about " lytic substance " is that,
as already mentioned, it may be derived from the bacteria
ab initio. Further, it is not necessary to wait for the chance of
its spontaneous emergence in an old culture. Various experimental
methods may be employed. For example, Otto and Winkler*
stated that lysin formation could be promoted in many different
ways, by adding to the culture filtrates from old cultures, by the
treatment of the bacteria with immune serum, and by physical or
chemical means (shaking in distilled water or the addition of a
small quantity of sublimate). But filtration through a bacterial
filter was especially useful and was preferable to centrifugalisation
because fewer passages were necessary in order to obtain the lysin.
Their method was to heat a broth culture at 58° C, filter, add a
few drops of filtrate to new broth containing fresh, living culture ,
and incubate for 24 hours at 37° C. These steps constituted the
first " passage." The process was repeated until the lysin was
obtained. Weinberg and Aznar j obtained similar results by the
filtration method with B. Shiga. They termed their products
" autobacteriolysins " because they were obtained directly and
primarily from the bacteria. They also showed, by the method of
autolysis in distilled water, that these lysins could be obtained
from young cultures as well as from old. They took two loopfuls
of a 24 hours' growth, emulsified it in 20 c.c. of sterile distilled
water, incubated at 37° C, and then demonstrated the presence
of lysin in the filtrate.

Lytic substances are of the same general characters, whether
produced directly with culture material or by more complicated
methods involving animal inoculation or the use of animal pro-
ducts. They are of bacterial origin but must be regarded as a
modification of bacterial constituents since they may not be
demonstrable as components of normal bacterial protoplasm. This
last fact may be illustrated by serological methods.

For example, Bordet and Ciuca J thoroughly immunised rabbits
with a normal strain of B. coli and found that the serum had no
antilytic properties. From the same strain, however, they obtained
a lytic principle and produced a lysogenic variant. The serum of
rabbits immunised with this variant was antilytic. Their inference
was that lytic power was undoubtedly a newly acquired character.

* Deutsch. med. Wochenschr., p. 383. 24th March, 1922.
t C.R. Soc, Biol. LXXXVI. (p. 833), 29th Apr., 1922, and LXXXVII.
(p. 136), 17th June, 1922.

% C.R. Soc. Biol, LXXXIV., p. 280, 29th Jan., 1921.

29

In the next place it has been shown that " lytic substances "
produce variants which are similar, in many important respects,
to those occurring naturally or spontaneously in cultures and
sometimes in the animal body. It is known, for example, that the
effect of the lytic agent upon bacterial growth is to cause a culture
to split up into two types, the " sensitive," which is highly sus-
ceptible to the action of this agent and rapidly undergoes autolysis
under its influence, and the " resistant," i.e., a type which resists
this action. Here reference may be made to Gratia's work* on
the production of variants with the aid of lytic substances.
Starting with a single strain of B. coli, he obtained eleven different
forms, each with distinctive characteristics though all still retained
the specific properties of B. coli. In the course of this work, he
says, " our attention has been called repeatedly to similar facts
reported by Arkwright." He refers to Arkwright's observations
on " rough " and " smooth " variants which are discussed above
by F. Griffith and me (p. 5 and p. 25).

Significance is to be attached to the fact that the initial
stimulus which produces transmissible " lytic substance " may be
derived from many quite different kinds of material, e.g., fsecal
extracts from healthy or diseased persons or from healthy animals
of various species, urine, diluted sewage, extracts of normal
organs, peritoneal exudates of guinea-pigs inoculated with some
bacterium, mixtures of bacterial culture and leucocytes, and so on.
It is of much interest to find that, whether the modification arises
" spontaneously " or under the influence of one or other of these
heterogeneous " lytic influences," the kinds of variants which are
produced are always very much the same. The reason, I think,
is that the modifying influence, whatever its nature, always acts
in the same sort of way, viz., by interfering with the normal
synthetic activities of bacterial protoplasm at the critical time
when the growing bacterium is about to subdivide.

Something may now be said about the transmissible properties
of lytic substance, which, since they are of bacterial origin, will
naturally depend on the characters of the bacterial protoplasm
from which they were derived.

Sometimes the individuality of the strain from which they
originated is marked very strongly. A good example is given by
Gratia. | He started with a lytic substance (bacterial filtrate)
which acted only on the particular strain of B. coli from which it
was derived. It was " without any action not only on other closely
" related species but also on other strains of Bacillus coli."

He found, however, that he could make it produce a lytic
substance possessing a wider range of action. From his original
strain of B. coli he separated out two types, the " sensitive " and
the relatively " resistant," upon both of which his first lytic
filtrate produced lysis. With the filtrate from the lysed
" resistant " strain, " a very marked dissolution was observed of

* Journ. Exper. Med., XXXV., p. 287, 1922.
t Journ. Exper. Med., XXIV., p. 115, 1921.

30

" Shiga bacilli, Flexner bacilli. Hiss Y type of Flexner bacilli, and
" also a strain of B. coli communis which was unattacked by the
" original filtrate." He further remarks that, in consequence of
these observations, which were made by Wollstein, " we have
" been able, by a method of successive passages through appro-
" priate strains, to extend the lytic power to other species, as
" typhoid and paratyphoid bacilli, and have obtained by this
" somewhat different technique results similar to those recently
" published by Bordet and Ciuca."

I think the point of special importance in the above observa-
tions is that the transition from the lytic principle (a), which
acts only on its own strain, to the wider lytic influence (6) is
effected without the introduction of any new factor. It is simply
the result of the continued action of a, which first produces rela-
tively resistant forms of the original strain and then, by attacking
and breaking down some of these resistant forms, gives rise to
the wider lytic principle 6. This means, expressed in terms of
the hypothesis which I am advocating, that the synthesis of the
bacterial protoplasm is interrupted more readily, or at an earlier
stage, in the case of the normal bacillus than in the case of the
resistant form ; with the latter, synthesis has gone a little further
before the lytic principle succeeds in producing its disintegrating
effect, and, consequently, the products of disintegration are
different.

From this point of view one may attempt to throw a little
light on the puzzling question, Why are there such remarkable
differences amongst what I may term " pure " lytic principles,
which are derived from different strains of the same bacterial
species ? By " pure " I mean that each is derived ab initio from
a particular strain of the species and has not been subjected to
any external modifying influence, such as passage through another
strain of the species or through a strain of another species. I
think the explanation, at least in part, is to be referred to the
extreme complexity of protoplasm, owing to which very slight
differences in its stability, or in the particular arrangement of its
constituent parts, at the critical phase of bacterial growth, may
result in disintegration products possessing different lytic
activities.

I think it is desirable first to recognise such possible com-
plexities of a single species, before embarking on still more
complicated problems such as (1) Why is a " coli " lytic principle
altered when it is transmitted through a culture of a different
species, such as Shiga, or (2) Does transmission of lytic principles
through different representatives of intestinal bacteria bring out
evidence of " group " relationship and different degrees of
kinship ?

Here is another interesting study which throws some light on
" lytic principles," because the experimental conditions are not
too complicated. It is concerned with the end results which are
obtained by continued exposure of bacteria to a lytic agent.

31

Martha Wollstein,* starting with the fact that the appropriate
"bacteriophage" for dysentery bacilli causes these organisms,
when plated out, to divide up into " regular " and " irregular "
colonies, compared the action of the lytic principle on cultures of
each of these types. It was found that the " regular " strains
were more resistant than the " irregular." Tests with the filtrates
from these strains showed that the lytic principle was carried by
the latter but not by the former. Both the regular and the
irregular colonies were subcultured on plates for more than
40 generations. The former retained their characters without
any alteration; the latter approximated to the former type,
until more regular than irregular colonies were present. " The
explanation seems to be that the sensitive bacilli die off more
rapidly than the resistant ones, which form the regular colonies
in later generations. It is a matter of selection." She then
endeavoured to ascertain whether the acquired lysogenic property
was retained permanently. " Working with the original, normal
culture as a whole, it was found that after the seventh generation
on agar or in broth the bacilli which had survived contact with a
lytic fluid were no longer able to transmit the lysogenic property
to other cultures or to dissolve the normal Shiga bacilli." As
regards the two main properties of the regular colonies, resistance
to lysis and loss of lysogenic power, it was found in some experi-
ments that a certain amount of lysis was produced on bacilli
cultured from regular colonies. The explanation was that these
cultures were not absolutely " pure," i.e., some less resistant
bacilli were still present. This was shown by re-plating, when both
regular and irregular colonies developed from a broth culture of
the strain in question, which had been in contact with lytic fluid
over night. The regular colonies picked from this plate were com-
pletely resistant to lytic action. Put briefly, the results of the
action of a lytic principle were : — (1) production of strains which
resisted lytic action, were not lysogenic, and were not agglutin-
able ; (2) production of strains which were lysogenic, agglutinable,
and sensitive to lytic action; (3) gradual conversion of these
latter strains into the resistant and non-lysogenic type.

If a lytic principle (a) derived from bacillus A is found to act
on a different bacillus B, it is not surprising that the lytic prin-
ciple (b) derived from filtrates of B may differ from a in its range
of action. But it is a curious fact (mentioned on p. 22) that b,
even after many transmissions through cultures of B, may retain
a certain impress of the individuality derived from A. This
individuality may be demonstrated by immunising animals with
a lytic substance and obtaining in their serum the corresponding
specific antibody. Two examples are worth quoting. Bruynoghe
and Appelmans| prepared antisera with two typhoid " bacterio-
phages," the one obtained from Strasburg and the other from
Louvain. The latter was originally a coli " bacteriophage " but

* Journ. Exper. Med., XXXIV., p. 467, 1921.

f C.R. Soe. Biol., LXXXVIL, p. 96, 27th May, 1921.

32

had been adapted to the typhoid bacillus and had lost all action
on coli. The antiserum obtained from the latter neutralised not
only the Lou vain " bacteriophage " but also its predecessor, the
coli " bacteriophage " ; it had no action on the Strasburg " bac-
teriophage." The Strasburg antiserum neutralised only the
Strasburg " bacteriophage." Gratia and Namur* obtained some-
what similar results with antisera for substances which were
lytic for staphylococci. The one substance (BH) was lytic
primarily for Staphylococcus aureus (H) and the other (BV) for
Staphylococcus albus (V) ; but, whereas BH acted on many strains
of these organisms, including V, BV was strictly specific for V.
The antiserum prepared with BH was specific for BH alone and
the antiserum for BV was equally selective for BV. After the
lytic principle BH had baen allowed to act on V and had been
transmitted ten times through V, the action of the two antisera
was tested on the lytic substance finally obtained. It was found
that no change had taken place ; specific neutralisation was still
obtained by the BH antiserum but not by the BV antiserum.

How do these antilytic sera act ? Do they behave like
ordinary antibodies which combine with specific constituents
corresponding to those of the antigen used for immunisation ?
Apparently they do ; at least, there is no reason to suppose that
they are an exception to the ordinary rule. In this connection
the following absorption experiments are worth noting. Jaumain
and Meulemanf placed in each of three tubes 5 c.c. of a lytic
principle active for staphylococci and added to tube (a) 10 drops
of broth, to (6) a like quantity of coli culture killed by heat, and
to (c) the same amount of staphylococci killed in the same way.
The tubes were sealed and kept in the incubator for three days.
Then the contents were tested quantitatively for lytic action on
staphylococci ; (a) and (b) had lost none of their activity, but
(c) was 1,000 times weaker. A few days later, filtrates from the
three tubes were tested ; (a) and (6) were as potent as the original ;
(c) was inactive. They stated that similar results, viz., absorption
of active principle by killed homologous bacteria, had been
obtained by Costa Cruz with B. coli. This absorption, they
found, was strictly specific and was not obtained if the killed
bacteria, though belonging to the same species, were derived from
a strain not sensitive to the lytic principle.

I mentioned above (p. 21) that it is convenient to draw a
distinction between the initial stimulus, which starts the produc-
tion of transmissible lytic material, and this material itself.
The former may be some non-specific chemical or physical
influence ; the latter is definitely a bacterial product. This
distinction is worth bearing in mind when one comes to consider
the significance of the lytic material which is obtained not from
the test-tube but from the animal body.

* C.R. Soc. Biol, LXXXVIL, p. 364. 24th June, 1^22.
f C.R. Soc. Biol, LXXXVIL, p. 362. 24th June, 1922.

33

Several observers have found that lytic material is present in
the serum of an animal shortly after inoculation with a small dose
of bacteria. To take a recently published example, Otto, Munter,
and Winkler* inoculated a guinea-pig intraperitoneally with
one-twentieth of a loopful of Flexner bacilli and found that lytic
principle was present in the serum obtained by heart puncture
seven hours afterwards. No lytic material was present in the
control (serum of the same guinea-pig obtained shortly before
inoculation with bacilli). How is the activity of the later sample
of serum to be explained ? Is it due to the presence of bacterial
products, derived from the Flexner bacilli ? Or does it mean that
the bacteria produced in the animal's plasma a temporary chemico-
physical change (Allergie), and that this changed condition,
being present in the serum, was operative in vitro as a primary
stimulus causing a growing culture to produce lytic substance ?

Similar phenomena have often been observed after injecting
animals with lysogenic bacterial extracts. Bordet and Ciuca.f
for example, injected a guinea-pig subcutaneously with 2 c.c. of
" lytic liquid " (a lysed and filtered suspension of B. coli) and
bled the animal seven hours afterwards. Twelve drops of the
serum, heated for half an hour at 56° C, prevented the growth of a
culture of B. coli, whereas the control serum, taken before inocula-
tion was ineffective. Similar results were obtained with the
serum of a rabbit taken from 7 to 24 hours after intravenous
injection with 20 c.c. of " lytic liquid." Here, again, the question
is whether the serum simply contained some of the original
lytic substance, which was still active, though highly diluted,
or whether the bacterial extract had modified the serum in a
way similar to that suggested at the end of the preceding paragraph.

In the same article Bordet and Ciuca raise a similar problem
about the transmission of antibodies. It is worth quoting, though
not strictly apposite to the present question (transmission of
lytic substance). After withdrawing a little blood from a guinea-
pig weighing 600 grammes, they inoculated the animal sub-
cutaneously with 5 c.c. of an antilytic serum and bled it 32 hours
subsequently. The two sera thus obtained were heated at
56° C, and then two drops of each were tested for antilytic
power ; the former serum was inactive, the latter active.

I am not prepared to offer any cut-and-dried explanation of
these phenomena. The main fact appears to be that bacterial
variants, associated with lytic change, are produced in the animal
body as well as in the test-tube. Whether the initial stimulus
which produced them is bacterial or animal, their propagation
depends on the bacteria themselves, i.e., on bacterial growth and
on the action of bacterial products. This is the point which I wish
to emphasise for the purpose of the subject now under discussion.
It would be beyond the scope of the present report to enter into
important collateral questions about the part played by the
animal body in resistance to bacterial infection.

* Zeitschr.f. Hyg., XCVL, p. 118, 1922.

f C.R. Soo. Biol, LXXXIV., p. 280. 29th Jan., 1922,

34

Summary.

It is now recognised that the study of bacterial variation is of
practical importance, because the subject is concerned with
variations in the capacity of bacteria for producing disease.

One is confronted with a confusing mass of data, ranging from
some minor and temporary change in a single function to the
much more profound alterations, popularly known as "the
Twort-d'Herelle phenomenon," where the vitality of the bacteria
is seriously imperilled and relatively few individuals survive.

It would simplify the matter if one could recognise some general
principle underlying this apparent confusion, and could attribute
the different types of variants to different phases in the operation
of this principle.

In the search for a common factor, it is observed that these
changes only occur with living and actually growing bacteria
and that, therefore, they are probably due to some influence
which operates at the nascent stage of growth. Hence the
common factor can only be explained in terms of vital processes ;
but about these no precise information is available.

Still, it is possible to formulate a general statement. The
constituents of living bacterial protoplasm have two functions,
catalytic action, i.e., the preparation of food material with the
aid of their appropriate enzymes, and synthetic action, i.e., the
building up of new protoplasm. It is obvious that the balance
between these two activities must be very minutely adjusted
and that any disturbance of this balance will tend to the pro-
duction of variants.

My report is concerned with the development of this theme
in relation to known facts about bacterial variants.

January, 1923.

Printed under the authority of His Majesty's Stationery Office

By Eyre and Spottiswoode, Ltd., East Harding Street, E.C 4,

Printers to the King's most Excellent Majesty.

PUBLIC HEALTH AND MEDICAL SUBJECTS.

The following Reports in this series have been issued by the
Ministry of Health since 1920 : —

No. 1. — The Complement Fixation Test in Syphilis, commonly

known as the Wassermann Test. (Price 5s.)
No. 2. — Incidence of Notifiable Infectious Diseases in each
Sanitary District in England and Wales during the
year 1919. (Price Is. 3c/.)
No. 3.— The Progress and Diffusion of Plague, Cholera and
Yellow Fever throughout the World, 1914-1917.
(Price 7s. 6c/.)
No. 4. — Report on the Pandemic of Influenza, 1918-1919.

(Price 10s.)
No. 5. — Dr. A. A. Jubb's Report to the Ministry of Health on
the Occurrence of Gastro-Intestinal Illness in the
Sandown Urban District, Isle of Wight. (Price 6c/.)
No. 6. — Incidence of Notifiable Infectious Diseases in each
Sanitary District in England and Wales during the
year 1920. (Price 2s.)
-Report on the Causation of Fcetal Deaths. (Price 10s.)
-Small-pox and Vaccination. (Price 3c/.)
-Diet in relation to Normal Nutrition. (Price 3c/.)
-Report on Diphtheria. (Price 9c/.)
-Report on Encephalitis Lethargica. (Price 10s.)
-Dr. W. V. Shaw's Report on an Outbreak of Enteric
Fever at Bolton-upon-Dearne. (Price 2s.)
No. 13. — Bacteriological Studies on the Pneumococcus. (Price

2s. 6d.)
No. 14. — Dr. T. W. Wade's Report on the occurrence of Bacillary

Dysentery in Glamorgan. (Price Is. 6cZ.)
No. 15. — Dr. Janet Campbell's Notes on the Teaching of
Obstetrics and Gynaecology in Medical Schools.
(Price Is. 3d.)
No. 16. — Report on Scarlet Fever and Diphtheria in Ramsbury.

(In press.)
No. 17. — Dr. J. M. Hamill's Notes on the Pasteurisation of Milk.

(In press.)
No. 19. — Report on the Use of Hydrocyanic Acid for Fumiga-
tion Purposes. (In press.)
No. 20. — Report on an Outbreak of Dysentery inLynton, Devon.
By G. C. Hancock, C.B.E. (In press.)

No.

7,

No.

8,

No.

9.

No.

10,

No.

11,

No.

12,

Copies of the above-named publications can be purchased
through any bookseller, or, directly from H.M. Stationery Office,
or at the following addresses : Imperial House, Kingsway,
London, W.C.2 ; 28, Abingdon Street, London, S.W.I; York
Street, Manchester; 1, St. Andrew's Crescent, Cardiff; or 120,
George Street, Edinburgh.

x 20179

LONDON:
PUBLISHED BY HIS MAJESTY'S STATIONERY OFFICE.

To be purchased through any Bookseller or directly from

H.M. STATIONERY OFFICE at the following addresses:

Imperial House, Kingsway, London, W.C. 2, and 28, Abingdon Street, London, S.W.

York Street, Manchester; 1, St. Andrew's Crescent, Cardiff;

or 120, George Street, Edinburgh.

1923.
Price Is. Od. Net.

I

QR
58
G74

Gt. Brit. Ministry of Health
Bacteriological studies

Biological
<& Medical

PLEASE DO NOT REMOVE
CARDS OR SLIPS FROM THIS POCKET

UNIVERSITY OF TORONTO LIBRARY

M ■■ ■

i