^^-m^^>^

J.

INJURY AND DEATH

OF

BACTERIA

BY

CHEMICAL AGENTS

OTTO RAHN

Professor of Bacteriology
CORNELL T NIVEKSITY

No. 3 of the "Biodynamica Monographs"
edited bv B. J. Luyet

I'ultlislied by

BI0DYNA:MICA, Normanuy, Missouri

1945

The contents of this monograph, except for a
few additions and some minor modifications, are
reprinted from Biodyxamica, No. 86, 1943 and No.
96, 1945.

FOREWORD

Tlie problem of the mechanism of death and that of
the mode of action of poisonous substances have been
among the favorite subjects of our program of publica-
tion. We, therefore, welcome the opportunity of editing
this monograph.

The author 's forty years of laboratory experience with
bacteria and his constant interest in the fundamental
problems of biology recommend him both for knowing
'^the facts" and for being one of those who try to ''get
to the bottom of things."

His notion that "cell mechanisms," corresponding to
the various physiological functions, are affected indi-
vidually by disinfectants, and that injury and death are
the result of an impairment or a destruction of some of
these mechanisms, will give rise, we expect, to fruitful
discussions.

New arguments are advanced, in the tirst part of the
monograph, in favor of the idea that the survivor curves
of bacteria — and, according to the author, of unicellu-
lar organisms in general — are exponential. There are
many opponents of this idea (among them the present
editor). Professor Rahn's arguments will force these
opponents to give thought to some aspects of the subject
which they might otherwise overlook.

The author's attempt to bring some system into the
study of the action of that group of substances which
stop, delay, or otherwise impede the activity of bacteria,
without killing them, will be of particular interest to
those who know how much of a "mix-up" this subject is.

June 1, 1945

Saint Louis, Missouri

B. LUYET.

TABLE OF CONTENTS

PAGE

FOREWORD 5

TABLE OF CONTENTS 7

INTRODUCTION _ i)

PART I

THE PROBLEM OF THE LOGARITHMIC
ORDER OF DEATH IN BACTERIA

I. The Order of Death 13

II. Order of Death in Multicellular Versus Unicel-

lular Organisms 20

III. The Cause of the Logarithmic Order of Death.. 28

The Deathrate Constant 28

Explanations of the Logarithmic Order 31

Mass Law and Single Molecules 37

IV. Conditions Resulting in a Non-Logarithmic Or-

der of Death 42

Cell Clusters and Chain Formation.. 42

The Effect of the Age of the Cell 42

The Order of Death in the Case of Low Death-
rates 47

The Order of Death of Other Unicellular Organ-
isms 49

Unexplained Deviations from the Logarithmic
Order 51

V. Biological Consequences of the Logarithmic Or-
der of Death 54

Importance of the Number of Individuals 54

Selection of Resistant Strains 56

Protection of Living by Dead Cells 58

PART II

MODE OF ACTION OF DISINFECTANTS AND
ANTISEPTICS

I. DISINFECTANTS

I. Selectivity of Disinfectants 63

II. Concentration and Death Rate... 66

59934

8 COXTEXTI^

PAGE

III. Temperature and Disinfection 74

Temiieratiue Coefficients of Life and Deatli 78

IV. Interference of Foreign Matter With Disinfec-

tion 80

V. Interference of Antidotes With Disinfection.... 82
VI. Properties of Various Groups of Disinfectants.. 84

Acids -. - 84

Alkalies - 85

Heavy Metal Compounds 85

Alcohols - 89

Formaldehyde 90

Oxidizing Agents 91

Halogens - 92

Soaps 95

VII. Standardization of Disinfectants 101

II. ANTISEPTICS

I. Temperature and Antisepsis 121

II. Selective Action of Antiseptics .' 123

III. Diluted Disinfectants as Antiseptics 128

IV. Effect of Size of Inoculum 133

V. The Dual Action of the Dyes 136

VI. Mode of Action of the Sulfonamides 146

VII. Increased Efficiency of the Weak Acids by In-
creased Acidity 157

VIII. Antagonistic Substances From Microorgan-
isms - - 164

SUMMARY AND CONCLISIONS 170

BIBLIOGRAPHY 175

SUBJECT INDEX _._ 179

AUTHOR INDEX _ 181

^y

V

INTRODUCTION

To understand the death of multicellular organisms, it
seems necessary to be acquainted with the death of indi-
vidual cells. This book deals with the death of single
cells, and although bacteria are not in all respects di-
rectly comparable with tissue cells, the fundamental
principles involved in death are the same. The study of
the death of bacteria furnishes, therefore, a solid founda-
tion for a general study of death.

Extensive researches on disinfection and sterilization
have resulted in a great deal of detailed information on
the death of bacteria. An attempt is made in this book to
sort out the experimental evidence and to search for the
basic reactions which cause death.

A study of death should logically begin with a defini-
tion of death, but right there, difficulties arise. Death
cannot be defined by positive criteria ; it can be character-
ized only by the absence of some property which is es-
sential to life. But different groups of biologists differ
in their views on the most essential properties of liv-
ing organisms. The death of a cell is not always de-
termined by the same method; the loss of motility, of
respiration or of other enzyme activities, the increase in
permeability which makes plasmolysis impossible and
causes the absorption of dyes, are variously used as
criteria of death. These criteria do not always appear
simultaneously (Rahn and Barnes, 1933) because, as will
be shown, they indicate an inactivation of different cellu-
lar mechanisms. Thus, a cell may be alive according to
one definition, and dead according to another.

The bacteriologist has no choice of definition because
bacteria are too small to permit an easy study of any of
the above criteria. The criterion almost universally used
by him to define death is the loss of reproduction. All
standard methods used in the study of disinfection and

10 INTRODUCTION

sterilization measure the death of bacteria by their in-
ability to produce colonies, or to make a culture medium
cloudy. It has been frequently overlooked in the general
study of death that, to the bacteriologist, a cell is dead
when it is sterile, i.e., peymauently unable to reproduce.
Sooner or later such cells will also lose the other, more
conspicuous properties characteristic of life. Although,
with larger animals, sterility of the organism is not iden-
tical with death, any organism in which the individual
cells have lost the ability to multiply will sooner or later
lose the other, more conspicuous properties of life.

The different criteria of death mentioned above refer
to the inactivation of different mechanisms which are
indispensable for the life and reproduction of the cell.
These mechanisms can be divided into five groups, ac-
cording to the functions they control:

Group I : Mechanisms controlling the passage of
substances into and out of the cell; they are repre-
sented by the cell membrane.

GroujD II : Mechanisms controlling the energy pro-
duction for all life activity ; they consist of enzymes.

Group III : The synthesis mechanism which prob-
ably involves a large number of different catalysts
that control the synthesis of the cell constituents
from the food.

Group IV : The multiplication mechanism : known
in higher organisms to have its seat in the nucleus.

Group V : Mechanisms consisting of protein mole-
cules which often make up the bulk of the cytoplasm,
but are without definitely known physiological func-
tion, though they are essential for the life of the cell
since their denaturation causes death. (Such are
myosin, leguminin, glutinin, the globulins of bacte-
ria, etc.)

IXTRODUCTION H

These mechanisms either consist of pure proteins, like
the chromosomes, or their major part is such, as in the
case of the membrane.

The first two and the hist of the mechanisms listed are
probably in<lependent of each other, while multiplication
cannot take place without synthesis, and synthesis can-
not take j)lace without energy, i.e., without enzyme action.

A partial or even complete inactivation of mechanisms
II or V need not cause death if the synthesis mechanism
can replace them. Synthesis has been observed, under
abnormal conditions, to continue while multiplication
was lost, resulting in single cells of very great length.
These cells are considered dead by the bacteriologist; in
fact, they disintegrate soon after they have ceased to in-
crease.

This consideration brings us to the problem of injury
to the cell. Boycott (1920) was led to believe from the
trend of the death curves obtained with bacteria that
they might be too simply organized to recover from in-
jury. How^ever, it seems that any organism that has the
ability to grow must have the ability to repair a certain
amount of injury. In multicellular organisms, recovery
is frequently not a repair of the injured cells, but is
accomplished through their replacement by new cells
which are produced by surrounding, uninjured tissues.
Similarly, in unicellular organisms, damaged protein
molecules may be replaced by new molecules produced
by the uninjured synthesis mechanism. Native proteins
are unstable, and the cell is continuously replacing inac-
tivated molecules. The same synthesis mechanism wiiich
originally formed all the cell constituents, and replaces
all mechanisms that deteriorate in the 'Svear and tear"
of life, may even work more rapidly in case of injury,
though within narrow limits.

A special kind of injury is of common occurrence in ^.^^-a^f^-^^^^
bacteria, namely, a temporary and reversible loss of re/\v51H?^^

.>."^'

^'library

^

12 INTRODUCTION

productive power. Such is the effect of antiseptics (as
opposed to that of disinfectants). The removal of the an-
tiseptic permits normal multiplication again. (In higher
organisms too, a reversible cessation of growth can be
brought about by aj^plication of toxic substances, but Ave
do not usually think of it as a parallel to antisej)sis.)

The course of death of bacteria is different from that
of all multicellular organisms, the death time being re-
lated logarithmically to the number of surviving bacteria.
This "logarifhmic order of death" is emphasized in Part
I of this monograjDh because it is the key to the under-
standing of the fundamental cause of death and of the
ultimate lethal reaction. The conclusions from Part I
are used in Part II to explain the cause of death by dis-
infecianfs. Besides, Part II shows the great differences
between the mode of action of disinfectants and that of
antiseptics.

PART I

THE PROBLEM OF THE LOGARITHMIC
ORDER OF DEATH IN BACTERIA

I. THE ORDER OF DEATH

Organisms which die as a result of a chemical reac-
tion die in an orderly, predictable way. Two methods of
graphic presentation of such orderly deaths are cus-
tomary. One consists in the use of mortality curves,
which are represented by black blocks in Figure 1, and
which give the number of individuals dying in each suc-
cessive time unit; this type of curve is commonly
used in medical statistics. In the other method use is
made of survivor curves, which show the number of indi-
viduals still living at any time after the beginning of ex-
posure. This type of curve, shown by the thin lines of
Fig. 1, is used by bacteriologists, entomologists and plant
pathologists.

The order of death is, in principle, the same for all
multicellular organisms. The mortality curve begins
at zero, and frequently remains at zero for some time,
as long as all individuals can recover after short ex-
posures ; then the first individuals die, the frequency of
death gradually increases to a maximum, then decreases,
usually at a slower rate, until only a few very resis-
tant organisms remain, and at last even they succumb.

Whether tadpoles, insects, or plant seeds are tested,
whether heat or rays or chemicals are used, the mor-
tality curve remains essentially the same. These mor-
tality and survivor curves agreed so well with our con-
ception of gradation in resistance that not much atten-
tion was paid to them. They were taken for granted until
Madsen and Nyman in 1907 and Harriet Chick in 1908

14

LOGARITHMIC ORDER OF DEATH

lOOV

TORULA MONOSA
AT 50*

50-

25

MOSQUITO LARVAE

IN Xheptylic acid

lOOV

SPORES OF

B. ANTHRACIS

IN MERCURIC CHLORIDE

JO'S

s o — o — i:

\flour beetle

75

\ X-RAYED

50

-

Y

29

■

^^^B^lBS ^*

lOOV

BACT. AERTRYKE
X-RAYED

Fig. 1. Mortality curves (black blocks) and survivor curves (thin
lines) of various organisms dying under the action of various killing
agents. Abscissa: time of exposure; ordinate: per cent of survivors.
(From data of Boycott, 1920, for tadpoles; of O'Kane, Westgate and
Clover, 1934, for mosquito larvae; of Davey, 1917, for flour beetle; of
Beamer and Tanner, 1939, for Torula; of Madsen and Nyman, 1907,
for B. anthracis; of Wyckoff, 1930, for B. aertryke.)

LOGARITHMIC ORDER OF DEATH

15

00%,

TORULA MONOSA
AT 50» C

5 ip 20 Minutat 30

1001
50

r \sO

SPORES OF E

k^NTHRACIS

10

r

x.^^^^^

5

^ IN MERCURIC

CHLOmbE^^

■

c

2 4

* 1

6 8 Min

1001
50

B.AERTRYKE

^^

10

r X- RATED

s

-

2 4 6 6

. ',".

12 Min.

Fig. 2. Semi-logarithmic plot of the survivor curves represented in
Fig. 1. Abscissa: time of exposure; ordinate: per ceiit of survivors
on logarithmic scale.

16 LOGARITHMIC ORDER OF DEATH

found that bacteria die in quite a different order. This
seemed so unusual and so significant that much experi-
mental work and many theoretical discussions have been
published to explain this interesting phenomenon. If the
culture for an experiment on the order of death is care-
fully chosen, not too young and not too old, the bacteria
die at a constant rate. This means that the same per-
centage of all bacteria alive at any given time will die iu
the next time unit. If, for instance, one-half of the bacteria
die in the first minute of exposure, one-half of the remain-
ing bacteria will die in the second minute. This leaves one-
fourth alive, of which again one-half will die during the
third minute and so forth. The survivors represent a geo-
metrical progression. If a is the original number, and s the
percentage of survivors, then the actual number of sur-
vivors in successive time units will be

a a-

100

vm) \-m)

If s = 10 (that is, if the survival rate is 10%) and a =
1,000,000 bacteria, the successive numbers of survivors
will be

1,000,000 100,000 10,000 1,000 100 10 1 0.1 0.01

A geometrical progression can be recognized graph-
ically by the straight line that one obtains by plotting
the logarithms of its members against their exponents.
Since, in a disinfection experiment, the number of sur-
viving bacteria plotted logarithmically against the time
of exposure, furnish a straight line, it has become cus-
tomary to speak of a ''logarithmic order of death" of
bacteria.

Figure 1 shows mortality curves and survivor curves.
The left side represents multicellular organisms, the
right side, bacteria. Figure 2 gives the corresponding
survivor curves in semi-logarithmic plotting. With bac-
teria, they are almost rectilinear ; with higher organisms,
they are concave downwards.

LOGARITHMIC ORDER OF DEATH 17

Table 1 gives other examples. The difference in the
order of death between bacteria and multicellular organ-
isms is here shown by the computation of the value

j^ _ _1_ |q„ initial number

t survivors

This value, the ''deathrate constant," is constant if the
order of death is logarithmic (see p. 30). The table
shows that, with bacteria, the successive values for K
either fluctuate around an average, or eventually de-
crease with time, while, with higher organisms, they in-
crease.

When no special effort is made to experiment with a
homogeneous material, the survivor curve^ may not be
straight but concave upwards. This is in fact the most
frequent occurrence. This type of curve differs from
that of higher organisms still more than the rectilinear
curve. Its concavity is due to lack of uniform resistance.
The more sensitive bacteria w^ill die rapidly, causing a
steep decline in the number of survivors. When most of
these bacteria are dead, the remaining ones die at a lower
rate, and as the less resistant individuals are gradually
removed, the survivor curve becomes less and less steep.

A simple, theoretical example is given in Table 2.
Three groups of bacteria, each containing 1,000 indi-
viduals, are supposed to be mixed and exposed to heat.
Each group represents a different degree of resistance,
the individuals of one group dying at the rate of 90%
per minute, those of the other two groups at 50% and
10%, respectively. The death rate of the composite
sample is not increasing as it would be with higher or-
ganisms of graded resistance, but it is decreasing. The
survivor curve (Fig. 3) is concave upwards, not down-
wards.

1. Unless otherwise indicated, the discussion of the characters
of survivor curves will always refer to semi-logarithmic plots.

18

LOGARITHMIC ORDER OF DEATH

TABLE 1

Mortality and death rate constants of different organisms dying
under the action of various killing agents.

(Data on fruit flies, by Loeb and Northrop, 1917; on mustard seeds,
by Hewlett, 1909; on flour beetle, by Davey, 1917; on Bad. coli (heat),
by Watkins, 1932; on Bact. paratiiDhosivm, by Chick, 1908; on Bad. coli
(light), by Clark and Gage, 1903).

Time of
exposure

Number

of
survivors

Mortality per
time interval

Indiv-
iduals

dead in

% of
survivors

Time of
exposure

Number

of
survivors

Mortality per
time interval

Indiv-
iduals

dead in

%of

survivors

Death by poisons

Death

by heat

Fruitflies exposed to 39.5°C

Bacterium coli exposed to 55 °C

0

100

0

0

6,000,000

25

95.5

0.9

0.9

o.ooos

5

1,110,000

4,890,000

81.3

0.147

30

83.0

12.5

13.1

0.0027

10

112,000

998,000

90.6

0.173

35

65.4

17.6

21.2

0.0053

15

51,300

60,700

54.3

0 138

40

45.4

20.0

30.6

0.0086

20

6,480

44,820

87.2

0.148

45

28.0

17.4

37.7

0.0123

25

998

5,482

84.5

0.151

50

17.0

10.9

3^.0

0.0154

30

457

541

54.2

0.137

55

11.8

5.3

31.2

0.0169

60

8.8

3.0

27.3

0.0176

65

2.5

5.3

60.3

0.0246

70

1.1

1.4

56.0

0.0279

75

0.3

0.8

72.7

0.0337

Mustard seeds in 0.2% HgCh

Bacterium paratyphosum in 0.6% phenol

0

1000

0

100

30

940

60

3.0

0.0009

1

86.1

13.9

13.9

0.065

45

895.8

44.2

4.7

0.0011

2

75.2

10.9

12.7

0.062

60

790.6

105.2

11.7

0.0017

3

64.4

10.8

14.3

0.060

75

486.6

304.0

38.6

0.0042

4

60.5

3.9

6.1

0.055

90

220.6

266.0

54.8

0.0078

6

49.6

6.0

9.9

0.051

105

163.8

56.8

25.8

0.0074

9

29.4

6.8

13.7

0.059

120

14G.0

17.8

11.0

0.0080

14

14.7

3.3

11.2

0.060

135

39.0

107.0

73.4

0.0104

19

9.2

1.1

7.5

0.055

31

3.7

0.5

5.5

0.046

Death by

irradiat

ion

Days

after

Flour beetle Tribolium exposed to

Bacterium coli exposed

to sunlight

jxposure

X-rays

(*)

1

0

100

0

71,400

1

98

2

2.0

15

12,365

59,035

82.6

0.051

2

98

0

0

30

3,568

8,797

71.2

0.043

3

97

1

1.0

45

1,065

2,503

68.3

0.041

4

93

4

4.1

(*)

60

285

770

72.6

0.040

5

72

21

22.6

6

43

29

40.2

7

17

26

60.5

8

3

14

82.5

9

1

2

66.7

10

0

1

100.0

*See footnote on p. 20.

LOGARITHMIC ORDER OF DEATH

19

TABLE 2

Order of death of three groups of bacteria representing three
different degrees of resistance, and of a composite group resulting from
a mixture of the former three (calculated values).

Number of survivors
in each group

Number of survivors
in the composite group

Death rate

90%

50%

10%

Total n. of
survivors

N. of surv.

per 1000

cells

Death
rate per
minute

0 minutes

1 minute

2 minutes

3 minutes

1000

100

10

1

1000
500
250
125

1000
900
810

728

3000

1500

1070

854

1000
500
357

285

50%'
29%
20%

Fig. 3. Order of death of three groups of bacteria representing three
different degrees of resistance (thin lines), and of a composite group
resulting from the mixture ol the former three (heavy line). Abscissa:
time of exposure; ordinate: number of survivors per 1000, on loga-
rithmic scale.

20 LOGARITHMIC ORDER OF DEATH

II. ORDER OF DEATH IN MULTICELLULAR VERSUS
UNICELLULAR ORGANISMS

The difference in the order of death between bacteria
and higher organisms has occasionally been questioned.*
A survey of all experiments on the order of death in
bacteria until 1930 (Rahn, 1931) showed that 25% of
the curves resembled those of multicellular organisms,
21% were strictly rectilinear, and 54% were concave
upward, like that of Figure 3. Most of the 25% re-
sembling multicellular organisms were actually obtained
with ''multicellular" bacteria, i.e., the bacteria were clus-
tered as will be explained in more detail on p. 42. Since
that survey a number of new investigations have veri-
fied the logarithmic order, e.g., the extensive investiga-
tions by Watkins and Winslow (1932), and by Beamer
and Tanner (1939a and 1939b). Only two authors, both
working with chlorine, have obtained survivor curves
which are consistently concave downwards. It cannot be
doubted any more that the majority of disinfectants,
as well as heat and irradiation, result in an order of
death in bacteria which is quite different from that of
higher animals and plants.

The difference between bacteria and higher organ-
isms which causes this difference in the order of death

*In the case of death by irradiation no mortality data comparable
to those obtained with other killing agents could be found in the
literature for higher organisms. The order of death is frequently given
in experiments with ultra-violet and X-rays, but the signification of
these data is different from that of other experiments. Irradiated
organisms appear normal for a long time after they have received a
lethal dose. The zero time for them is the moment when irradiation
ceased, while, in all other experiments, it is the moment when exposure
begins. If an insect, for example, does not die until a week after it
received the lethal dose, it is clear that time-mortality curves mean
very little.

This point is important in the study of the ultimate mechanism of
death. Evidently, If a plant or animal survives for days after it has
received a lethal dose, it is not the destruction of its enzymes and
cytoplasm which is responsible for its death. Yet, the organism dies.
Some irreparable damage has been done to some cell function which
is not immediately needed. The similarity of this phenomenon Avith
the fact that irradiated gonads often produce normal-appearing animals
which, however, are sterile, suggests that death by irradiation is due
to the inactivation of some genes.

LOGARITHMIC ORDER OF DEATH 21

is a most obvious one : with bacteria, the individual is
dead when a single cell dies ; with higher organisms, the
death of one cell does not kill the organism. It is not
known how many cells of any tissue must be killed to
cause, e.g., the death of a tadpole or an insect or a plant
seed, but it is certain that no single cell is so all-im-
portant that its death causes the death of the entire
organism.

The order of death of the individual cells in the
tissues of higher plants or animals would be very diffi-
cult to ascertain because in most cases it is impossible
to obtain cells of equal resistance and to expose them
simultaneously. But it seems reasonable that the single
cells of tissues should die like the single free-living cells.

Let us now investigate the order of death by heat in
a multicellular organism, such as a tadpole, assuming
that there is in that animal a vital organ, for example,
the brain, which is most sensitive to heat. The survivor
curve of the individual brain cells would be

("iw") ""(w) ■■■ Ki5r)

"* ''Tor "

The probability that a certain definite cell in the brain
is still alive after t time units is

p=(T5(r)'

and the probability that it is dead is
P = l-

100

Since s/100 is smaller than 1, (s/100)' decreases rap-
idly with longer exposure times. Therefore, the prob-
ability that this one cell is dead increases rapidly with
prolonged exposure.

The probability of death of another definite brain cell
is the same. The probability that both these cells are
dead at the time t is

22

LOGARITHMIC ORDER OF DEATH

and the probability that n cells are dead is

p»=[i-(w)'T

If the tadpole is dead when n brain cells are dead,

Pn

expresses the probability of its death, or the fraction of
the total number of animals which is dead at the time t.
If we multiply this probability b}^ 100, we obtain the per-
centage of dead tadpoles at the time t.

The entire difference of order of death is explained
by this equation. Table 3 shows, on the basis of cal-
culated values, how the mortality and the order of
death vary when the number of cells whose inactivation

TABLE 3
Mortality (number dying per minute) in a population of 100
multicellular organisms, calculated on the assumption that an individ-
ual dies when n of its cells are inactivated.

500

If 90% of the cells die per minute (the survival rate being 10%)

1

2

5

10

50

100

1st minute

90.00

81.00

59.05

34.87

0.52

0

0

2nd minute

9.00

17.01

36.05

55.56

59.98

36.60

0.64

3rd minute

0.90

1.79

4.40

8.57

34.62

52.78

59.96

4th minute

0.09

0.18

0.45

0.90

4.38

9.62

34.57

5th minute

0.01

0.02

0.05

0.09

0.45

0.90

4.33

If 80% of the cells die per minute (the survival rate being 20%)

1st minute .
2nd minute.
3rd minute
4th minute.
5th minute
6th minute.
7th minute.

80.00

64.00

32.77

10.74

0

0

16.00

28.16

48.77

55.75

13.00

1.69

3.20

6.24

14.52

25.79

53.93

43.11

0.64

1.28

3.14

6.13

25.38

40.42

0.13

0.28

0.64

1.27

6.12

11.66

0.02

0.03

0.13

0.26

1.26

2.49

0.01

0.01

0.02

0.05

0.25

0.50

0

0

1.80
43.13
40.42
11.50

2.52

If 70% of the cells die per minute (the survival rate being 30%)

1st minute .
2nd minute.
3rd minute.
4th minute.
5th minute.
6th minute.
7th minute.
8th minute.
9th minute.

70.00
21.00
6.30
1.89
0.56
0.17
0.05
0.02
0.01

49.00
33.81
11.87
3.72
1.12
0.33
0.11
0.03
0.01

16.81
45.60
24.79
8.82
2.77
0.84
0.26
0.08
0.02

2.82

36.12

37.12

16.13

5.41

1.67

0.51

0.16

0.04

0

0.90

24.51

41.19

21.94

7.87

2.50

0.77

0.22

0
0

6.48

37.87

34.05

14.55

4.87

1.55

0.43

0

0

0

1.71
27.91
39.73
20.20

7.28

2.15

LOGARITHMIC ORDER OF DEATH

23

causes the death of the organism varies from 1 to 500.
The calculation is carried out for survival rates of 10%,
20% and 30%. Table 4 shows the gradual decrease in
the number of survivors, for a survival rate of 50%.
In Table 6 the percentages of survivors are compared
when the death of n definite, or that of n random cells
kills the organism.

Figure 4 gives the mortality curves and Fig. 5, the
survivor curves on semi-logarithmic scale, for various
values of n, and for a survival rate of 50%. When n =
1, the mortality curve as well as the survivor curve are
plainly those of bacteria. As n increases, the curves be-
come more and more similar to those characteristic of
higher organisms, and for n = 8 or more, they are prac-
tically identical with them.

Thus, the two different orders do not involve dif-
ferences in the chemical dynamics of death; they are
merely different results of the same principle at dif-
ferent levels of organization. As death of one single
cell does not kill a large animal or plant, the organ-
ism survives until enough cells are inactivated to cause

TABLE 4
Number of survivors in a population of 100 muIticeUular organ-
isms, calculated on the assumption that an individual dies when n of
its cells are inactivated, and that the death rate of the cells is 50%
per minute.

n =

1

2

4

* 8

16

32

64

128

1000

Minutes:

1

50.00

75.00

93.75

99.61

100.00

100.00

100.00

100.00

100.00

2

25.00

43.75

68.61

89.99

99.00

99.99

100.00

100.00

100.00

3

12.50

23.44

41.63

65.64

88.19

98.61

99.98

100.00

100.00

4

6.25

12.11

23.00

40.33

64.39

87.32

98.39

99.97

100.00

5

3.12

6.15

12.18

22.43

39.83

63.80

86.89

98.28

100.00

6

1.56

3.10

6.10

11.84

22.27

39.59

63.50

86.68

100.00

7

0.78

1..56

3.09

6.08

11.80

22.20

39.47

63.37

99.96

8

0.39

0.78

1.55

3.08

6.07

11.77

22.16

39.41

98.00

9

0.20

0.39

0.79

1.55

3.08

6.06

11.76

22.14

85.84

10

0.10

0.20

0.39

0.78

1.55

3.08

6.06

11.77

62.40

11

0.05

0.10

0.20

0.39

0.78

1.55

3.07

6.06

38.60

12

0.02

0.05

0.10

0.19

0.39

0.77

1.54

3.05

21.40

13

0.01

0.03

0.05

0.10

0.19

0.39

0.77

1.54

11.40

14

0

0.01

0.03

0.05

0.10

0.19

0.39

0.77

6.00

15

0

0

0.01

0.03

0.05

0.10

0.20

0.39

3.02

24

LOGARITHMIC ORDER OF DEATH

death. The different order is merely due to the fact that
higher organisms are multicellular.

These tables and figures disclose another point which,
if not very obvious, is quite important. The assumption
made at the beginning of the calculations was that the
cells die in logarithmic order. This implies that they
are of uniform resistance. On the other hand, there can
hardly be any doubt that, in any large population, all

8 MIN.

Fig. 4. Mortality curves (deaths per minute) of multicellular organ-
isms the death of which is caused by the inactivation of n cells.

100%

14 MINUTES

Fig. 5. Semi-logarithmic plot of the survivor curves of multicellular
organisms the death of which is brought about by the destruction of n
cells. Abscissa: time of exposure; ordinate: per cent of survivors on
a logarithmic scale.

LOGARITHMIC ORDER OF DEATH

25

the cells are not of uniform resistance. Our assumption,
then, seems to disagree with the facts. There is no
contradiction, however, since the assumption does not
deny the actual existence of a graded resistance, it merely
states what is the consequence of a uniform resistance.
But there is a point where the theory of the loga-
rithmic order of death is in direct contradiction with the
theory which explains the mortality curves by a graded
resistance. If the typical mortality curve of higher or-
ganisms, which is spread over a certain length of time,
were due to a gradation in resistance, a uniform resis-
tance should result in crowding the death of all indi-
viduals into the same instant. This last point is pre-
cisely what is contradicted by the theory of the logarith-
mic order.

To illustrate this point, let us assume a population of
tadpoles killed by heat, where death occurs when n =
100 brain cells are incapacitated. Three grades of re-
sistance are assumed to be present, characterized by sur-
vival rates of 10%, 20% and 30%, respectively; the pop-
ulation consists of 100 individuals of each grade. Table
3 and Table 5 give the mortality for each grade and Table
5 gives also the mortality for the entire population. Fig-
ure 6 illustrates the fact that tadpoles of the same re-
sistance do not die at the same moment, and that the
maximum mortality occurs later and later, as the re-
sistance increases. The composite mortality curve for

TABLE 5
Calculated mortality curve of a population consisting of organisms
of three groups of different resistance.

Number

of

Individuals

Group I: 100
Group II: 100
Group III: 100
Total: 300
Percentage

Survival
rate per
minute

Number of individuals dying per minute

26

LOGARITHMIC ORDER OF DEATH

the total population is still a regular mortality curve,
drawn out over a longer time than the curves of each
grade. If the time scale is reduced to one-half, as in
Figure 6a, (which gives the mortalities : 12.7, 63.4, 21.2, 2.4
and 0.1 for successive 2-minute intervals) the shape of
the curve is the typical one. Curves and Tables show
that it is not only the sensitive cells that die early, and
not only the resistant ones that survive for the longest
time ; in both groups there is a small percentage of cells
of different degrees of resistance.

Fig. 6a

Fig. 6. Mortality curves of three groups of organisms representing
different grades of resistance (thin lines) and of a population con-
sisting of equal numbers from each of the three groups (heavy line).

Fig. 6a. Curve of the composite population of Fig. 6 on a reduced
time scale.

TABLE 6
Number of survivors in a population of 100 organisms, calculated
on the assumption that the death of n out of 10 cells kills one organism.
Comparison of n definite versus n random cells. The survival rate
of the cells is assumed to be 90% per minute.

In the case of n definite cells

Minutes

n = l

n=2

n=3

n=4

n=5

n = 6

n=7

n = 8

n = 9

1

90

99.0

99.9

100

100

100

100

100

100

2

81

96.4

99.3

99.9

100

100

100

100

100

3

72.9

92.2

97.8

99.4

99.9

100

100

100

100

4

65.6

86.3

94.9

98.1

99.3

99.8

99.1

100

100

5

59.0

78.8

90.2

95.5

97.9

99.1

99.5

99.8

99.9

In

the case of n

random cells

1

0

55

88

97.9

100

100

100

100

100

2

0

0

18

71.8

93.7

99.0

100

100

100

3

0

0

0

0

56.6

89.9

98.4

99.8

100

4

0

0

0

0

0

46.0

88.5

98.4

99.9

5

0

0

0

• 0

0

0

42.5

90.9

99.0

LOGARITHMIC ORDER OF DEATH 27

The above calculation has been made on the assump-
tion that we were dealing- with ''definite" cells, while
death of the multicellular organism will result from the
destruction of any ''random" cells. The inactivation of
n definite cells will take much more time than the inacti-
vation of n random cells. The correction factor which
changes the formula for definite to that for random
cells is

a!

(a-n)! n!

The probability that any n cells out of a total of a cells
are inactivated, is then (see p. 21)

P ' =

a!

(a - n) ! n !

1-

\ 100 j _

With a— 10 cells, the probability that any one of these
10 cells dies is 10 times greater than the probability that
a definite cell, e.g., Number 3, dies. The probability of
death for any two cells is

10-9-8-7-6-5-4-3-2-1

8-7-6-5-4-3-2-lx2-l

or 45 times as great as for any two definite cells. Table
6 gives the number of survivors in the case of definite
and of random cells for a survival rate of 90%.
Generally speaking, since

(a-n) !n !

is a constant, it cannot alter the shape of the survivor
curves or of the mortality curves. The rate is greatly
increased, but only by a constant factor; the convex
curves remain convex, and the straight lines remain
straight. If more than one cell must become inactivated
to cause the death of an organism, the order of death is
not logarithmic, Avhether these cells are definite ones or
not.

28 LOGARITHMIC ORDER OF DEATH

III. THE CAUSE OF THE LOGARITHMIC ORDER OF DEATH

T h e D e a t h r a t e Co n s t a n t . The outstanding
feature of the death of bacteria is the constancy of the
deathrate. "Constant deathrate" means that the num-
ber of bacteria which die per time unit is a constant per-
centage of the number of living cells at the beginning of
this time unit. (One may notice that the number of bac-
teria dying in each of the successive time units decreases
continuously when the percentage in question remains
constant.) The constancy of the death rate leads to the
geometrical progression mentioned above (p. 16) for the
successive number of survivors :

^ ''Tor ""Kiw) \~im~) '" ^\ 100

where a is the initial number and s the percentage of
survivors. After t time units, the number of survivors is

b =a

\ 100 /

or

t log-^^^ = log b- log a

In this equation, s is constant, and therefore log s/100
may be called Ci; a is also constant, and log a may be
called C2 ; ?> is variable and is a function of the exposure
time t. Thus, we have

logb = Cit + C2

which means that the logarithms of the numbers of
survivors plotted against the time of exposure lie on a
straight line.

LOGARITHMK' ORDER OF DEATH 29

It is inconvenient to express the rate of dying- as the
percentage of survivors or deaths per time unit since the
time intervals may vary even within one experiment. It
has become customary to compute a constant, the ''death
rate constant," which is analogous to the reaction con-
stant of chemical reactions. If, at the time t, x bacteria
out of a initial bacteria have died, the number of sur-
vivors is a-x, and the rate of death is

= k (a - x)

which, upon integration, gives

a-x

The constancy of k in this formula is frequently used as
a criterion for the logarithmic order. But, for the pur-
pose of testing the constancy of the k values, it is cus-
tomary to use decimal logarithms instead of natural log-
arithms as the equation requires. This gives rise to a
new constant, which will be designated by K.

K = 0.434k = A_log initial number
t survivors

Tables 1 and 7 show that K (and therefore k) in-
creases with time in the case of multicellular organisms,
but is constant or decreases in the case of bacteria.

The percentages of survivors per time unit can be
found from the equation

K = logJ^

log s = 2 - K

This substitution of decimal for natural logarithms
may not be permissible if the constants are to be used
for further mathematical calculations.

30

LOGARITHMIC ORDER OF DEATH

The following table shows the relation between the
deathrate constant K and the percentage of bacteria
killed per time unit, which is (100-s).

If 10%

are ki

lied

per time

unit, 1

\r =

0.045

20%

' :=

0.096*

30%

' =

0.153

40%

' =

0.222

50%

' =

0.301

60%

' =

0.398

70%

' z=z

0.523

80%

' =

0.699

90%

' =

1.0

95%

' =

1.301

99%

' =

2.0

99.9%

' =

3.0

Sandholzer and Katzin (1942) found a simple way of
expressing the logarithmic relation in practical appli-
cations. They point out that, for a survival ratio of 10%,

Kt = log.

a

and therefore

0.1a

t =

= log 10 = 1

K

Then, instead of speaking of deathrates or percentage
of survivors, they use as basis of all their calculations
this "decimal reduction time," i.e., the time necessary to
reduce the bacterial population to one-tenth of its orig-
inal value. It is obvious that a doubling of this time
will reduce the number of bacteria to 1% of the initial
number, and trebling would leave only 1 out of 1,000 cells
alive. This terminology is applied to the pasteurization
of milk. The decimal reduction time is more readily vis-
ualized than the death rate. Generally speaking, death
time is more commonly used in applied sterilization and
disinfection than death rate.

LOGARITHMIC ORDER OF DEATH 31

Explanations of the Logarithmic
Order. The striking difference between the mortality
curves of bacteria and those of higher organisms has at-
tracted the attention of many biologists. According to
their attitude towards this problem, they can be divided
into three groups. The first group denies the existence
of a logarithmic order of death. They maintain that
bacteria have the same type of survivor curve as higher
organisms, but that the rapidity of bacterial death makes
us overlook a first period of low mortality, which, though
very short, would nevertheless be present. The second
group accepts the logarithmic order as an experimental
fact and claims that it is brought about by an unusual
distribution of bacterial resistance; various biological
reasons for this peculiar gradation of resistance are of-
fered. The third group believes that variation in re-
sistance is a minor issue, and that the logarithmic order
must be explained by analogy to the monomolecular re-
actions of chemistry.

The attitude of the first group can be characterized by
the following quotation from Loeb and Northrop (1917) :
''Miss Chick states that ... in each interval of time the
same percentage of individuals alive at this time is
killed. She was probably led to this assumption by the
fact that the ascending branch of the mortality curve in
her experiments was generally very steep. The agencies
used by her for killing the bacteria were so powerful that
the ascending branch became almost a vertical line, thus
escaping attention. Hence she noticed usually only the
less steep descending branch which could be interpreted
as a monomolecular curve for the reason that her exper-
iments lasted only a short time."

Knaysi (1930, 1-V) took up this general line of thought
in five short papers. We are primarily concerned here
with his experiments with weak disinfectants (paper II)
which show a very slow decrease in the number of or-
ganisms for the first two minutes, and which resulted in

32 LOGARJTHMIC ORDER OF DEATH

curves that he considers sigmoid. Six of the experi-
ments required 180 minutes ; the shortest one lasted 52
minutes. This represents a slow death for disinfection
experiments. It will be shown on p. 48 that a sigmoid
survivor curve must be expected when death is so slow
that repair becomes appreciable.

After the first few minutes, Knaysi's curves are shaped
quite differently from any survivor curves for higher
organisms (plotted on semi-logarithmic scale). Five are
concave upwards, indicating inhomogeneous material
(probably due to the use of cells from agar cultures in-
stead of from the customary liquid cultures), two are
straight, and three are so irregular that they would not
fit any theory. Knaysi concluded that "these results
can be adequately explained only if the distribution of
resistance in cultures of bacteria is considered to govern
the course of the process."

Knaysi later (paper V) approached the problem math-
ematically by using Pearson's curves of variability. The
curves shown in his graphs are quite ditferent from any
mortality curves known for higher organisms. He fur-
ther quotes Yule who "demonstrated that if the cells are
to be considered all alike, and if the action of poison is
noncumulative, the law of chance gives an exponential sur-
vivors' curve. If the action is cumulative, the death rate
constantly increases." However, in Knaysi's experi-
ments, the death rate increases only for a few minutes
and then declines rapidly.

But the main point of the controversy is the shape
of the mortality curves during the very first minutes of
exposure. The determination of this shape is practi-
cally impossible because one cannot prove that, with
bacteria, death begins at the moment of exposure, and
because one can always argue that the bacteria which
were found dead after the first minute, all died in the
second half of this minute. Even for the x-ray experi-
ment of Figure 8 where the first observation was made

LOGARITHMIC ORDER OF DEATH 33

after 6 seconds, it could be claimed that no cell died
in the first 3 seconds. Though such a claim may be
entirely unjustified, it is obviously impossible to disprove
it.

But those who profess a constant death rate in bac-
teria can disprove the identity of the two orders of death
by pointing to the survivor curves. If the mortality
curves of bacteria and those of higher organisms are es-
sentially similar, the semilogarithmic graphs of the sur-
vivor curves of higher organisms should be straight lines
when the first two or three survivor counts are omitted.
A glance at Figure 2 will show that this is not the case.

Far more helpful in one's decision as to the nature of
the curve is the death rate constant K, because each
value of K is significant in itself independently of the
previous or of the subsequent trend of the curve. It has
been shown that, with higher organisms, (or with clus-
tered bacteria, or also in the case of very slow death, as
will be explained below) K is not constant, but increases
with time; that, with uniform bacteria, it is constant;
and that, with bacteria of unequal resistance {e.g., with
very young cultures), it decreases consistently. To illus-
trate these differences of death rate, a few character-
istic data (where K is multiplied by 1,000 for easier com-
parison) have been gathered in Table 7. They represent
many of the organisms commonly used in experiments
on death. Each set gives the K-values for all successive
time units for which survivors were counted.

Subsequent to the objections of Loeb and Northrop,
of Knaysi and of many others, Watkins and Winslow
(1932) made a very extensive study of the order of
death of Bacterium coll killed by heat and by dilute
NaOH. (For the complete 150 sets of data, see Watkins,
1932). They concluded: "The studies . . . seem to indi-
cate that when vegetative cells {Esch. coll) are exposed
to the toxic action of sodium hydroxide, reduction pro-
ceeds at a generally logarithmic rate and does not ex-

34

LOGARITHMIC ORDER OF DEATH

TABLE 7

Death rate constants at successive time intervals for various
organisms killed by heat or by poisons. (The original values of the
death rate constants K are multiplied by 1000.)

Higher
organisms

Bacteria and Yeasts

Clustered

Single, uniform

Single, mixed

Death by heat

' ' i

Sacchar.

Bacter.

Bacter.

Fruit-

Tadpoles

Wheat

Htai

)h. Microc.

Bacter.

ellips-

para-

para-

Strept.

Flies

Seeds '

aurc

us sulfureus

coli

oideus

typhosum

typhosum

lactis

0

0

0

If

5 103

82

32

63

748

4431

0

2

0 •

r

6 177

83

30

68

355

5210

0

25 ■

0

2:

3 139

87

36

76

222

4495

0

67

0

3?

0 267

102

37

77

166

3335

4

150

0

484

85

39

76

133

2412

13

0

458

89

30

74

113

2229

26

17

405

34

68

107

1851

43

27

J ■

336

32

65

61

46

407

65

77

95

84

117

88

141

123

140

Age

Age

167

55 hours

5 hours

Loeband

Boycott,

Groves,

(^hi

ck, Sattler,

Chick,

Beamer

Reichen-

Reichen-

Sattler,

Northrop,

1920

1917

19

0 1928

1910

and

bach,

bach.

1928

1917

Tanner,
1938

1911

1911

Death by poisons

Tomato

Mustard

Mosquito

moth

Bacillus

Bacillus

.\nthrax

Bacter.

Bacter.

Bacter.

Bacter.

seeds,

larvae.

larvae,

No. 25,

X,

.spores.

coli.

coli.

typhosum.

paratyph

HgCl2

Heptylic
acid

Lead
arsenate

NaOH

HgClo

HgCb

NaOH

CaCl2

Phenol

Phenol

0

0

45

0 58

11

146

13

256

60

62

1.8

0

73

0.58

74

126

9

220

67

56

2.3

0

143

0

01

70

138

10

237

52

58

5 2

0 3

171

0

68

82

149

9

213

59

46

8.7

0 8

214

0

75

107

204

11

57

38

8.7

1.5

472

0

86

106

186

13

57

33

8.0

2 3

584

08

116

147

11

.

58

32

11.7

3.6
4.7
5.3
6.7
8.9
8.3

901
2470

11

57
55
56
56

30
23

22
20

Hewlett,

O'Kane

Lloyd,

Mvers,

Reich-

Madscn

Watk'ns,

Winsiow

Lee and

Chick,

1909

et al..

1920

1929

enbach.

and

1932

and Falk,

Gilbert,

1908

1934

1911

Nyman,
1907

1926

1922

liibit the increasing mortality rate with the progress of
time which has been observed by Levine and his asso-
ciates for spores. In our experiments, values for K (the
coefficient for the rate of reduction per unit of time)

LOGARITHMIC ORDER OF DEATH 35

were approximately constant throughout the experi-
ment. We have estimated the ratio of K for the two
time-halves of an experiment and find this ratio to aver-
age 0.97 for 109 experiments with NaOH, 1.17 for 16
experiments with a borate-buffer mixture and 0.96 for
26 experiments with hot water (50° and 55°)."

The second group of explanations comes from biolo-
gists who accept the logarithmic order of death as an ex-
perimental fact and try to explain how the resistance is
graded in a different way in bacteria and higher organ-
isms. Eeichenbach (1911) and Henderson Smith (1921)
are the only authors who seem to have presented really
original ideas on the subject.

Eeichenbach assumed that, in a bacterial culture, a
definite proportion of cells from each generation ceases
to multiply and becomes increasingly more resistant dur-
ing the ensuing rest period. Such a culture would ex-
hibit an exponentially graded resistance. This assump-
tion, it must be recognized, has no experimental foun-
dation. However, it would lead to the same result to as-
sume that the bacterial cell does not divide into two equal
parts, but that the mother cell produces a daughter cell
as in the case of yeasts, and that the mother cell is defi-
nitely older and more resistant than the daughter cell.
This mode of multiplication would result in an expo-
nentially graded resistance which under certain condi-
tions could yield a perfect logarithmic order. But the
theory, aside from its unverified morphological impli-
cations, fails to explain the logarithmic order of death
in the case of spores. It seems impossible that the as-
sumed type of gradation of resistance could occur in the
case of spores.

Henderson Smith (1921) points out that a variation in
resistance may be brought about by differences in the
thickness of the cell membranes. Penetration by poison
may be proportional to the square of the thickness, and
this would result in an order of death approximating the

36 LOaARITBM/V ORDER OF DEATH

logarithmic. This explanation to which one may object
again that it is entirely theoretical and not based upon
any measurements of membrane thickness (see however
Figure 12) — might explain chemical disinfection, but
it does not seem plausible for heat sterilization, and can
not explain the order of death in the case of X-rays
which penetrate with lightning speed.

Lee and Gilbert (1918) call attention to the fact that
biological characteristics are as a rule distributed in a
manner quite different from that which has been as-
sumed in the various theories of graded resistance. "In
view of these observations, the authors are led to the
conclusion that the logarithmic nature of the disinfec-
tion process is due to a general similarity of the indi-
viduals in a given pure culture rather than to a dissim-
ilarity of the individuals."

The third group of explanations is based on the analogy
with monomolecular chemical reactions. The all-inclusive
statement by Paul (1909) that bacteria are so small that
they react chemically like large colloidal molecules
will probably not be taken seriously by the biologists of
today.

More in line with present conceptions is the explana-
tion proposed by Isaacs (1935) who elaborated on the
contention of many biologists that enzyme inactivation
is probably the first step in the death of the cell. The
fact that enzyme inactivation is frequently a monomole-
cular process led Isaacs to the hasty conclusion that the
logarithmic order of enzyme inactivation is the evident
cause of the logarithmic order of death of bacteria.

This conclusion is based on a wrong application of the
law of mass action. When an enzyme solution, distrib-
uted in 100 test tubes, is heated, and one of these tubes
shows 90% of inactivated molecules, we are certain that
all other test tubes at that moment also contain 90% of
inactivated molecules. The order of enzyme inactiva-
tion in each test tube is logarithmic, and the reaction

LOGARITHMIC ORDER OF DEATH 37

must necessarily be identical in all tubes, as long as
the mass law applies. If, in identical test tubes, the re-
action would proceed dilTerently, there could be no mass
law and no science of chemistry. Thus, if bacteria would
die from inactivation of enzymes, all cells should die at
the same moment.

The mass law makes only one assumption, namely,
the presence of masses {i.e., of large numbers) of re-
acting molecules. This requirement is fulfilled in our
case. Estimates by Hand (1933) for catalase in yeast
suggest about 2,000 molecules per cell. Haldane and
Stern (1932) concluded that the yeast cell contains be-
tween 15,000 and 150,000 invertase molecules. The as-
sumption of fairly large numbers of enzyme molecules in
bacteria, which are in many respects comparable with
yeast, seems thus justified, and the logarithmic order
of enzyme inactivation cannot explain the logarithmic
order of death of bacteria.

Mass L a IV and Single Molecules.
However, if the number of molecules becomes very small,
identical cells would not die at the same time. The mass
law can not be applied when there is no mass. When a
million test tubes, each containing a solution of 1,000 en-
zyme molecules, are heated uniformly, it is quite cer-
tain that when half of the molecules are inactivated in one
tube, the same fraction will be inactivated in all the other
tubes. When the same solution is put in much smaller
test tubes so that each contains only 10 enzyme mole-
cules, it is to be expected that, while 5 molecules are in-
activated in most of the tubes, there may be 4 or 6 inacti-
vated in some of them. If we choose our tubes so small
that each contains only one enzyme molecule, the reac-
tion cannot be identical in all the tubes. Yet, the mass
law still holds for the entire enzyme solution. By chang-
ing the size of the container, the rate of reaction can-
not be altered, and, at the same time and temperature,
half of the molecules must be inactivated in these small

38

LOGARITHMIC ORDER OF DEATH

tubes. This means that, of these uniform test tubes under
uniform conditions, one-half contains inactivated mole-
cules and the other does not. It is possible to compute
from the mass law the number of tubes in which a single
protein molecule is inactivated, but it must be realized
that the result represents only a probability and not a
certainty. The certainty of the mass law is founded on
the very large number of molecules present in ordinary
chemical reactions. In living cells, not all kinds of mole-
cules are so numerous.

If p% of the molecules are inactivated per minute, the
total of a tubes will be divided into those with an in-
activated, and those with an unchanged molecule, as
follows :

Time

Tubes with
unchanged
^ molecules

Tubes with inactivated
molecules

P

a

a

100
P

100

P
100

all
P II

a

100
P

P \ P
100/ 100

)

_P_
100.

a

V 100/

['-(:

100/

LOGARITHMIC ORDER OF DEATH

39

The probability that a certain tube (or molecule) has
undergone a change at the time t is

Px= 1

1--P-V

100/

This formula for the probability of inactivation ap-
plies to any one definite enzyme molecule, whether in
a separate small test tube or in the original large volume
of enzyme solution. The probability that tmo definite
molecules, either in the large volume or in the small test
tube, be inactivated is the product of the probabilities
of each separate event:

Pa = Pi -Pi =

1-1

P

100

and the probability that n molecules be inactivated is

P. =

P

100

As was stated above, p is the percentage of molecules
inactivated per minute. The percentage not changed is
100 — p = s. If we introduce this into the above equa-
tion, we obtain

R =

100

This is the same formula which we encountered on page
22. All calculations in Tables 3 and 4, and the curves
of Figures 4 and 5 apply to this case ; we merely have to
substitute the word ''molecule" for "cell."

The conclusion is that a logarithmic order of death can
be obtained onl}^ if the death of the bacterium is brought
about by the reaction of one single molecule. This con-
clusion is absolute. The logarithmic order of death is
entirely impossible if more than one molecule must be in-
activated to produce the death of the cell. As Tables 3
and 4 and Figure 4 show, an approximation of the loga-
rithmic order is obtained if death is caused by inactiva-

40 LOGARITHMIC ORDER OF DEATH

tion of a very few molecules but if this number is higher
than 4 or 5, the order is very definitely not logarith-
mic. This eliminates denaturation of enzymes or of the
cytoplasm as the cause of death, wherever the logarith-
mic order prevails, because it does not seem possible that
inactivation of a very few of the numerous identical mole-
cules can kill the organism. It would probably exclude
also the disruption of the cell membrane as a cause of
death, for this is likely to require inactivation of a con-
siderable number of molecules before damage becomes ir-
reparable.

Let us now consider the bacteriologist's definition of
death : a bacterium is dead when it has lost the power to
reproduce. Cell division is linked with chromsome di-
vision, the chromsomes contain the genes, and the inac-
tivation of certain genes is known to cause ''lethal muta-
tions," i.e., to prevent the cell from multiplying. While
chromsomes and genes have not been definitely found in
bacteria, these organisms undoubtedly have some heredi-
tary mechanism, and hereditary units. As a rule, there
are two genes of the same kind in a diploid cell, and one
in a haploid cell. According to Fricke and Demerec
(1937): "we may assume that a gene contains about
2500 atoms . . . This would indicate an average gene di-
ameter of about 25 A." This is the size of a quite small
protein molecule. A gene, then, would consist of only
one or two molecules. Each gene is ditferent from all
others, and has its own rate of denaturation. If one vi-
tally essential gene is denatured, and the cell can no
longer divide, the bacterium is sterile, that is, dead, ac-
cording to the bacteriological definition.

This explanation, namely, that ''death" of bacteria is
brought about by the inactivation of a gene (Rahn, 1929,
1934) is correct only for the bacteriologist's definition
of death. Jordan (1940) defined the death of bac-
teria as "lethal mutation." The very extensive data on
mutations brought about by X-rays, and on the lethal ef-

LOGARTTHMTC ORDER OF DEATH 41

fects of X-rays, all point very definitely to the fact that
the genes are the most sensitive part of the cell. Fre-
quently, while no other effects of irradiation are immedi-
ately noticeable on the irradiated plants or animals, ster-
ility has been induced, or, when the affected genes are
less essential, the offspring is abnormal in some way.

To this mass of indirect evidence has recently been
added the chemical production of mutants in fungi. Thom
and Steinberg (1939) obtained stable mutants of several
Aspcrgilli by making the spores germinate on agar
containing KNO,. All mutations consisted in the loss
of some activity related to the reproductive function, and
have been explained by the authors as being caused by
inactivation of some amino groups in one of the genes.

Analogous findings have been recorded in bacteria.
Penfold (1911) grew Bacterium coli on agar containing
chloracetate ; among the few bacteria that survived, were
some which had permanently lost the power to produce
gas from glucose. Eevis (1912) obtained the same re-
sult with malachite green and brilliant green. Lom-
mel (1926), by means of weak disinfectants, produced a
variant of Bacterium coli which had lost the ability to fer-
ment lactose. Unpigmented "albino" strains of Bacte-
rium prodigiosum can be obtained by cultivation at very
high temperature. All the mutations cited were irrever-
sible. If, in bacteria and fungi, weak disinfectants can
permanently inactivate some genes, which control fer-
mentation, reproduction or pigmentation, it is highly
probable that, in general, they cause changes in genes nec-
essary for life and thereby produce sterile mutants,
which the bacteriologist would consider dead.

42 LOGARITHMIC ORDER OF DEATH

IV. CONDITIONS RESULTING IN A NON-LOGARITHMIC
ORDER OF DEATH

Cell Clusters a n d Chain Formation.
The logarithmic order of death of bacteria can be estab-
lished only by counting the survivors, and the only prac-
tical method in use is the plate count method which gives
the number of colonies developing from a known volume
of bacterial suspension. The number of colonies repre-
sents the number of original bacteria only when the bac-
teria are single. If they are clustered like Staphylococ-
cus, or in long chains like Bacillus mycoides or some
Streptococcus strains, one colony may represent a large
number of cells. As long as not all cells from such a clus-
ter die, the rest will still give rise to a colony, and death
becomes evident only when the last of the cells in that
cluster is dead. Such clusters *'die" like multicellular
organisms, and their survivor curves must be concave
downwards. This is commonly observed with Staphylo-
coccus (see Fig. 7).

Substantial support was given to this explanation by
Wyckoff and Rivers (1930) who irradiated bacteria
spread of an agar surface (see Fig. 8). In one experi-
ment the bacteria were exposed immediately after hav-
ing been spread ; the survivor curve was then nearly rec-
tilinear. In another experiment the bacteria remained
several hours on the agar surface before irradiation,
and had multiplied to aggregates of perhaps 4 to 16
cells. As long as not every cell in such an aggregate
was killed by X-rays, a colony developed, and the sur-
vivor curve was similar to that of a multicellular organ-
ism. Many recorded bacterial survivor curves resem-
bling those of higher organisms can be interpreted in this
way (Rahn, 1930).

The Effect of the Age of the Cell.
It has been demonstrated by Sherman and Albus (1923)

LOGARITHMIC ORDER OF DEATH

43

Fig. 7. Semi-logarithmic plot of survivor curves of Staphylococcus
aureus killed by 0.6% phenol. (From data of Chick, 1908; the two
curves have been drawn to the same scale.)

that young bacterial cells succumb more readily to harm-
ful influences than old cells. The term "old cells" is ap-
plied to cells of cultures which have nearly or completely
reached the maximal population. With the customary
technique of the bacteriologist and with the customary
test cultures, this stage is reached in about 24 hours.
*' Young cells" are cells from cultures 2 to 8 hours old.
An increase of the age beyond that required to reach the
maximal population does not usually increase the resis-
tance of the cells very much. Although many cells lose
their viability after the stage of maximal population is
passed, those cells which remain viable do not seem to be
weakened by old age, and their resistance to disinfect-
ants or heat is not decreased. However, they require a
longer time for adjustment before they can start to mul-
tiply again in a new medium.

44

LO(h\RlTHMI<' ORDER OF DEATH

When old cells are transferred to a new environment
they are dormant and do not all begin to grow at the same
time. Some cells start much earlier than others, and the
general picture is probably similar to that of the germin-
ation of Aspergillus spores shown in Figure 12. This
leads to great contrasts in sensitivity since very young
and very old cells occur side by side. If such cultures
are treated with antiseptics or heat, the survivor curves
are very markedly concave upwards. They become more
and more nearly straight as the culture gets older. This

Time of irradiation in seconds

Fig. 8. Order of death of Bacterium coli killed by X-rays. The solid
line represents the order of death when the bacteria were exposed to
X-rays immediately after being spread on agar; the dotted line, when
they were exposed a few hours after being spread. (From data of
Wyckoff and Rivers, 1930.)

LOGARITHMIC ORDER OF DEATH

45

was shown by Eeichenbach (1911) with a culture of Bac-
terium paratyphosum exposed to heat (see Fig. 9; since
the temperatures used by this author were not exactly
alike in the different experiments, the data have been
made comparable by recalculation to a standard scale
where 99.9% of the inoculum are killed in 100 time units.)

2 0

60 TIME. UNITS .
— 1, ■ — ^ •-

Fig. 9. Semi-logarithmic plot of survivor curves of a culture of
Bacterium paratyiihosmn exposed to heat at different time intervals
after inoculation. Abscissa: time of exposure. (From data of Reich-
enbach, 1911.)

The observation that young cells are more sensitive
than old ones is readily accepted by biologists because it
agrees with analogous findings with higher plants and
animals. Even in higher organisms, however, certain ex-
ceptions must be acknowledged, for instance, the greater
resistance of children against certain diseases. It is in-
teresting that similar exceptions have been observed by
Hoffmann (unpublished) with bacteria. Gentian violet
iias a much stronger effect on old bacteria than on young
ones.

46

lAXSAUlTHMlC ORDER OF DEATH

Fig. 10. Sharp breaks in survivor curves.
Left: Bacterium 8chottmuelleri killed by exposure to 5.5°.
Right: Monilia Candida killed by exposure to 50°.
(From data of Beamer and Tanner, 1939a, b.)

Another phenomenon probably due to the effect of
age upon the resistance of bacteria, although definite
proof is still lacking, is the abrupt break in survivor
curves which has been frequently observed in heat disin-
fection, but which occurs also in chemical disinfection.
Figure 10 shows a typical example.

The first quantitative records of this phenomenon were
made by Gage and Stoughton (1906) who found that
when old cultures of Bacterium coli were exposed to
heat, a few cells remained alive for a long time after the
large majority had died. In one experiment, 99.99% of
the cells were killed in 5 minutes at 60^, but a few sur-
vived even when heated to 85° for 5 minutes.

The great practical difficulties arising from this phe-
nomenon in pasteurization and sterilization will be dis-

LOGARITHMIC ORDER OF DEATH 47

cussed elsewhere, but the principle should be discussed
here. In those instances in which this outstanding resis-
tance of a few individuals was oberved, the number of
resistant cells was always very small, usually much small-
er than 0.1%. Even if these cells were morphologically
different, it would be difficult to find them under the
microscope among the mass of normal cells. But it is
probable that they are not different, for their progeny
differs in no way from the parent culture (not even in
resistance, as will be seen below).

It may be that the few survivors are very old cells
which failed to come out of their resting stage when
transferred to a new medium. In general, one loopful of
a grown culture transfers between one million and ten
million cells. If some of these remain in the resting
stage for 24 hours, a new transfer from this culture
would carry over a considerable number of cells much
older than the average, and they would not multiply
readily, but would more probably remain in the rest-
ing stage. It would require at least three successive
transfers at short intervals to eliminate such cells. This
explanation does not require that the resistance increase
with age. It merely assumes that all other cells are much
younger and therefore more sensitive.

The Or d e r of Death in the Case of
Low Death rates. When death is very slow, the
survivor curves seem to be generally concave downwards.
However, most of Knaysi's experiments (1930, II) with
very weak disinfectants show a decreasing death rate and
survivor curves which are concave upwards. This indi-
cates some inhomogeneity in the material, which was
probably due to the use of agar-grown bacteria instead of
broth-grown cells. Woerz (1931, unpublished) observed
that Torula cremoris, which does not form clusters, died
in logarithmic order when held at a temperature well
above the maximal temperature of growth; but at tem-

48

LOGARITHMIC ORDER OF DEATH

peratures near this maximum, the survivor curve be-
came definitely concave downwards (Fig. 11). If we
realize that with many species the synthesis mechanism
can reproduce all the parts of a cell in 15 to 30 minutes,
it is not surprising that the effect of repair becomes
perceptible when the process of dying is extended over
several hours.

The following simplified analysis might make clear
the writer's conception of the interaction between death
and repair. Let us assume that death is brought about
by the inactivation of the mechanism of cell division and
that the earliest phase of this process can be reversed by
the synthesis mechanism. Let the rate of injury be 50%
(that is, affect 50% of the cells) per hour, and the rate
of repair during the first hour be 40%. Then, only 10%
of the cells cannot be repaired, and the death rate is 10%.
But, the harmful environment affects not only reproduc-
tion, but also the synthesis mechanism which, we as-

Fig. 11. Death by heat of Torula cremoris. This graph shows that
different types of curves may be obtained when death is slow (at 48°)
and when it is rapid (at 49.1°). From data of Woerz, 1931, unpub-
lished.)

LOGARITHMIC ORDER OF DEATH

49

TABLE 8

Dependance of the death rate on the rates of injury and of repaii*.
Each figure represents the per cent of the cells which survived at the
end of the previous time unit.

At given rates of injury and

of repair

one has

after

1 hour

2 hours

3 hours

4 hours

5 hours

Cells injured every hour. . . .
Cells repaired every hour . . .
Cells dying every hour

50%
40%
10%

50%
20%
30%

50%
10%
40%

50%

5%
45%

50%

2.5%
47.5%

When the process of injury is accelerated 10 times while the
decrease of the repair rate remains the same, the corresponding
figures are:

after

6 min.

12 min.

18 min. 24 min.

30 min.

Cells injured every 6 min . . .
Cells repaired every 6 min. .
Cells dying every 6 min ....

50%

4%

46%

50%

2%
48%

50%

1%
49%

50%
0.5%
49.5%

50%
0.25%
49.75%

sume, loses half of its efficiency per hour. Thus, repair
during the second hour is only 20% while injury con-
tinues its constant rate of 50%, which results in the death
of 50-20 = 30% during the second hour. In the third
hour, with repair decreased to 10%, 40% die. The fur-
ther stages are shown in Table 8.

The second half of this table shows what happens if
death is much more rapid. It is assumed that injury
has become 10 times as fast, so that 50% of all cells are
now injured in 6 minutes instead of in 60 minutes. The
amount of repair possible in this short time is almost
negligible, even if we assume that injury of the synthesis
mechanism is not increased, and the percentage of cells
dying per time unit is now nearly constant from the be-
ginning.

The Order of Death of Other Uni-
cellular Organisms. If death, i.e., loss of re-

50 LOGARITHMIC ORDER OF DEATH

production, is due to the lethal inactivation of some geue,
this must apply not only to bacteria, but to all other
unicellular beings as well, at least as long as they are mo-
nonucleate. That the order of death of yeast is logarith-
mic has been shown by Woerz (Fig. 11), by Rahn and
Barnes (1933) and especially by Beamer and Tanner
(1939b). Some of the data obtained by the last mentioned
authors are given in Fig. 2 and Table 7. Since certain
species of yeast have a tendency to form cell aggregates
in consequence of the fact that the daughter cells, and
even the third generations, frequently remain attached to
the mother cells, the survivor curves may eventually be-
come concave downwards (Eijkman, 1912). Beamer and
Tanner avoided clumps by shaking their stock cultures
for 30 minutes and filtering them through cotton before
exposure.

The survivor curves of mold spores, whether death is
caused by heat or by disinfectants, are concave down-
wards (Henderson Smith, 1921 and 1923). Although mul-
tinuclear, the spores could not develop, or would at least
become abnormal, if one of the nuclei were inactivated.
On that basis one would expect a logarithmic order of
death. But it is probable that spores are fairly resistant
in their natural dry state, and the penetration of water
is known to be a slow process requiring considerable
time. Germination is often irregular. Fig. 12 shows
the rate of germination of the spores of Aspergillus niger
on an agar surface. The percentage of germination was
ascertained every 10 minutes, and the mold spores were
found to be very inhomogeneous. Since germination as
well as disinfection depends above all upon the penetra-
tion of water, such material cannot be expected to show a
constant death rate. However, X-rays should cause a
logarithmic order since they penetrate instantaneously.

LOGARITHMIC ORDER OF DEATH

51

in

u

2 rn

■

s

3
t

f-.o

a.

J

■i^H

^d!

1

■^^tlS^JlBK.

^■■3{

□^■■a

.^ r1

Fig. 12. Irregularity in the number of spores of Aspergillus which
germinate per time unit (on an agar surface). Abscissa: time from
the moistening of the spores; ordinate: per cent of germinating spores.

The order of death of algae has been studied occasion-
ally. Harvey (1909) found a logarithmic order when
Chlamydomonas was treated with hydrochloric acid (Fig.
13). The criterion was the loss of motility.

With protozoa, the same criterion resulted in a sur-
vivor curve which was pronouncedly concave down-
wards. Peters (1920) studied death in the flagellate Col-
pidium in dilute solutions of Hg Cl> ; the survivor curve
he obtained is shown in Figure 14. It is possible, of course,
that loss of motility is one of the latest symptoms to be-
come noticeable, and that the protozoa were doomed to
die before this criterion could be observed.

Unexplained Deviations from the
Logarithmic Order. The evidence on hand is
overwhelmingly in favor of logarithmic order of death in
bacteria, and isolated reports of exceptions could be dis-
regarded. However, there are some consistant excep-
tions which cannot be disregarded although they are un-
explained. The following two very extensive sets of ex-
periments on disinfection by chlorine, carried out inde-

52

LOGARITHMIC ORDER OF DEATH

15 MINUTES 25

Fig. 13. Semi-logarithmic plot of the survivor curve obtained vi^ith
the alga Chlmnydomonas killed by 0.009% HCl. Abscissa: time of
exposure; ordinate: number of survivors on logarithmic scale. (From
data of Harvey, 1909.)

""0'~~-^^

■

O^v^^^

V)

.

^

o

>

.

">

%-

3

•

-

\p

H-

1.0

o

v.

(A

\>

£

\

j=

\

^—

\

k.'

•

\

a

\

o»

\

o.

\

J

0

\

3

1 1 1

6

1

9 Minutes \
■ 1 1 i —

Fig. 14. Semi-logarithmic plot of the survivor curve obtained with
the protozoon Colpidiuvi killed by mercury bichloride. (From data of
Peters, 1920.)

LOGARITHMIC ORDER OF DEATH

53

pendently, resulted in survivor curves which were con-
cave downwards. Charlton and Levine (1937) suspend-
ed the dried spores of a Bacillus (either laterosporus or
r^uminatus) —which, were found by microscopic inspec-
tion to be mostly single — in water with varying amounts
of chlorine, and counted the survivors by plating. Mall-
mann and Ardrey (1940) used a suspension of Bacterium
coli in water which had been chlorinated 30 minutes pre-
viously; their results are represented in Fig. 15.

It is remarkable that chlorine differs so much in this
respect from most other disinfectants. No comparison
with iodine could be made since the order of death by
iodine does not seem to have been investigated. It is con-
ceivable that chlorine, being such a very active chemical
reagent, kills bacteria by destroying the membrane, or
the protoplasm, or the enzymes before attacking the
mechanism of reproduction. This would result in the
observed order of death.

Fig. 15. Semi-logarithmic plot of survivor curves of Bacterium coli
treated with chlorine. Abscissa: time of exposure; ordinate: per cent
of survivors, on logarithmic scale. (From data of Mallmann and
Audrey, 1940.)

54 LOGARITHMIC ORDER OF DEATH

V. BIOLOGICAL CONSEQUENCES OF THE LOGARITHMIC
ORDER OF DEATH

Importance of the Number of I n -
dividuals. Let us consider again the case discussed
on page 16 of a million bacteria per cubic centimeter
dying at the rate of 90% per minute. The survivors at 9
consecutive minutes will be :

100,000 10,000 1,000 100 10 1 0.1 0.01 0.001

(This last number means that there is one bacterium left
alive in 1,000 c.c. If the bacterial suspension were dis-
tributed in ampules of 10 c.c. each, one ampule out of a
hundred would still contain one living bacterium.) If
instead of a million, we had only 1,000 bacteria per c.c.
at the outset, the same degree of sterilization would
have been reached in 6 minutes instead of in 9. The expo-
sure time for sterilization thus depends upon the amount
of contamination. It is a logarithmic function of the
number of cells present. If we have two suspensions
of the same bacterium, one with a cells and the other
with ma, and if the death times required to reach the
same degree of sterilization in the two cases are, respec-
tively, ti and to, we have the equations

Kt, = log ^

Kt2 = log

a-x
ma

ma-y

X and y being the number of bacteria dead at the times,
t, and to, respectively. The same degree of sterilization is
reached when a-x=m-y. Subtracting the second equation
from the first we obtain :

K(t2-t,)=logm
This means that if one inoculum is m times as large as
the other, the increase in death time (to-ti) is propor-
tional to log m, and inversely proportional to the rate
of death.

LOGARITHMIC ORDER OF DEATH

55

This equation could be used, in the following form, to
determine the death rate :

K =

log m

That the amount of contamination influences very defi-
nitely the time required for sterilization, is an every-day
experience in the dairy industry, in the canning indus-
try and in surgery. Quantitative data, given by Esty and
Myers (1922) on spores of Clostridium botuJinum, are
reproduced in Table 9. Sometimes K is not constant,

TABLE 9
Time required to sterilize by heat cultures of different concentra-
tions of spores of Clostridium botulinum. (From Esty and Meyer
1922.) '

Number of
spores
per cc.

Some spores

still
alive after

All spores
dead
after

log m
K-

t2-ti

900,000,000

9,000,000

90,000

900

9

44 minutes
34 minutes
18 minutes
12 minutes

48 minutes
36 minutes
20 minutes
14 minutes
2 minutes

105°

0.17
0.14
0.17
0 17

72,000,000,000

1,640,000,000

32,000,000

650,000

16,400

328

230 minutes

120 minutes

105 minutes

80 minutes

45 minutes

35 minutes

240 minutes

125 minutes

110 minutes

85 minutes

50 minutes

40 minutes

100°

0.013
0.026
0.032
0.035
0.041

but decreases as the cell concentration increases. This
is attributable to the protective effect of large numbers of
dead cells (see p. 59). The second half of Table 9 shows
this effect when the number of spores is nearly 100 times
as large as that employed in the upper half.

The outstanding fact that an increase in infection in-
creases the death time is contrary to any explanation of
the order of death by various degrees of resistance. Cells
of very high resistance are rare and a large quantity of
bacteria may contain a few extremely resistant cells
which are not usually encountered in small quantities.

56 LOGARITHMIC ORDER OF DEATH

But, as Hastings, Fred and Carroll (1925) remark, "one
million spores should include all grades of heat resis-
tance." Then in the 900 million suspension of Table 9
there should be no spore which is more resistant than
those in the 9 million suspension. Accordingly, though
the suspension with 900,000,000 spores survives 12 min-
utes longer than the one with 9,000,000, this difference
cannot be explained by a graded resistance. The assump-
tion that death is caused by the inactivation of a single
molecule per cell fits the facts perfectly.

This influence of very large numbers of individuals
should make itself felt also in higher organisms. In
Table 4, for instance, for n = 4, it was computed that
after 5 minutes, 1 out of 10 individuals is alive, after 8
minutes, 1 out of 100, after 12 minutes 1 out of 1,000 and
after 15 minutes 1 out of 10,000. As n increases, the ab-
solute time differences remain the same, although their
effect is relatively smaller. Actual experiments with
very large numbers of higher organisms are not known
to the author. Possibly some investigations on the via-
bility of seeds have been made which would give infor-
mation on this point.

8 el e ct i 0 n \o f Resistant Strains. Bac-
teria can become accustomed, to a certain degree, to ad-
verse conditions, and strains of remarkable resistance
do occasionally occur. If the last survivor is the most
resistant, it should be easy to produce such a resistant
strain by simply propagating the last survivors from a
large number of exposed bacteria. However, if the death
of the cell results from a lethal mutation, the last sur-
vivor is not different from the bacterium which is the
first to die (see page 40).

Several attempts to obtain resistant strains from the
last survivors have been published, and in most cases,
failure was reported. One of the most extensive and
persistent attempts is that by Gage and Stoughton (1906)
who tried to obtain a heat-resistant strain of Bacterium

LOGARITHMIC ORDER OF DEATH

57

coli by cultivating- the last survivors of a culture heated
to 60° C. Resistance was measured by the percentage of
survivors after heating 5 minutes to temperatures from
45° to 100°. The survivors of the 60° heating were again
cultivated, and tested, and this was carried through for
nine ''generations." (See the results on Table 10). Sum-
marizing, the authors write: "Experiment 195 in which
we attempted to produce by the survival of the fittest a

TABLE 10
Comparative resistance of successive cultures of last survivors
of B. coli. Each figure represents the per cent of survivors after 5
minutes heating at the temperature indicated. (From Gage and Stough-
ton, 1906.) B. coli can multiply at 45°.

Generations

Heated,

°C.

1st

2nd

3rd

4th

5th

6th

7th

8th

9th

45

136.0%

86.0%

100.0%

63.0%

72.0%

63.0%

258.0%

64.0%

243.0

50

70.0

3.6

44.0

46.0

20.0

23.0

4.5

9.7

49.0

55

4.4

0.006

0.050

2.3

0.004

0.020

0.020

0.012

0.200

60

0.001

0.008

0.010

0.001

0.009

0

0.160

0.002

0

65

0.250

0.400

0.001

0.001

+

0

0.020

0

0

70

0.030

0.001

0.002

0

0

0

0.010

0

0.022

75

0.050

0.014

0.003

+

0

0

0

0

0

80

0.015

+

0.003

+

0

0

0

0

0.018

85

0.003

0

+

0

0

0

0

0

0

90

0.003

0

0

0

0

0

0

0

0

95

0

0

0

0

0

0

0

0

0

100

0

0

0

0

0

0

0

0

0

race of especially resistant organisms, demonstrated a
diametrically opposite result." This decrease in resis-
tance by rejDcated transfer of last survivors is shown in
the Table.

Gates (1929) found that "cocci from colonies of the
last surviving organisms have proved to be inher-
ently no more than normally resistant to ultraviolet
light."

The positive results obtained by Magoon (1926) are
probably due to an erroneous interpretation of the data
(Rahn, 1932, page 299).

Strains of unusual resistance can be obtained by ex-
posure of large numbers of bacteria and by continuous

58

LOGARITHMIC ORDER OF DEATH

transfer of large numbers to media containing different
amounts of disinfectants. The customary means of trans-
fer is a loop which holds about 0.01 cc. of liquid. One
loopful of a grown culture of bacteria contains be-
tween 1 and 10 million bacteria. Only a few of these
can adapt themselves to the disinfectant, and only these
cells multiply. This does not mean that they would be
the last survivors if stronger concentrations were ap-
plied. In this way, Borman (1932) obtained strains of
Bacterium cjU resistant to salts of Fe, Cu, and Hg,
and Header and Feirer (1926) could greatly increase
the resistance of several species of bacteria to silver
nitrate, mercurochrome, formaldehyde, acriflavin, hexyl
resorcinol and phenol.

The outstanding cases of acquired drug-fastness are
met with in chemotherapy where a large number of path-
ogenic micro-organisms are exposed to the drugs.

Protection of Living h y Dead C ells.
Lange (1922) was probably the first investigator to dem-
onstrate experimentally that the presence of dead bacte-
ria retards the death of others. He found that the death

TABLE 11

Protection of living cells by dead cells in broth. (From Lange,
1922.)

Standard suspension

Water

Broth

Culture living (-I-) or dead (0)
after

Living cells Dead cells

5' 10'

20'

30' 45' 60'

120'

180'

^40'

Bact. coli at 60°C.

0.02 drops .

Ice

5 cc
5 cc

+
+

4-
+

+

+
+

0

+

0

+

0
0

0
0

0

0.02 drops .

1 cc.

0

Micr. pyogenes at 56°C.

0.02 drops .

1 cc

5 cc
5 cc

+
+

+
+

+
+

0

+

0

+

0

+

0
0

0
0

0

0.02 drops .

1 cc.

0

LOGARITHMIC ORDER OF DEATH

59

time of Bacterium coli or Staphylococcus aureus was dou-
bled or even tripled by the addition of dead cells of the
same species, provided that the cells were young when
killed. Old cells gave no protection, (cf. Table 11).
"Watkins and Winslow (1932) observed the same effect
in disinfection by alkali as well as by heat (Table 12).
Such decrease in the death rate by accumulation of dead
cells would tend to make the survivor curve slightly con-
cave upwards, and this is the most commonly observed
shape.

TABLE 12
Decrease of the death rate constant corresponding to an increase
in cell concentration. (Data of Watkins and Winslow, 1932.)

Killing agent

Temper-
ature

Initial cell concentration in millions per cc.

over 100

50-99

1-49

less than 1

n/200 NaOH....
n/150 NaOH...
n/100 NaOH... .
n/125 NaOH... .
n/100 NaOH...
Borate Buffer . . .
Heat

30°
30°
30°

22°
22°
30°
50°
55°

0.015
0.055
0.128
0.051
0.095
0.280
0.003
0.088

0.018
0.112
0.217
0.062
0.122
0.323

0.082

0.045
0.150
0.465
0.120
0.222
0.402
0.006
0.118

0.061

Heat

No satisfactory explanation has been given which cov-
ers all cases. It is obviously impossible to assume an
equilibrium between dead and living cells correponding
to the equilibria of chemical reactions. Watkins and
Winslow state: "We can only suggest the possibility that
a zone of protective substances in the menstruum sur-
rounding a given cell may react upon the chemical or
physical condition of the cell wall itself in such a way
as to make it more resistant to the influence of heat."

Recently this problem has come to the foreground
again in the study of the efficiency of the sulfonamide
compounds. It is usually stated that amino benzoic acid
(which neutralizes the effect of sulfa compounds) is re-
leased in small quantities from autolyzing cells. This

60 LOaARTTHMIC ORDER OF DEATH

explanation, which will be discussed elsewhere fits only
the case of this one group of disinfectants and cannot
be applied to other types of disinfectants or to death
by heat.

PART II

MODE OF ACTION OF DISINFECTANTS
AND ANTISEPTICS

In the fight against bacteria, a rather surprising va-
riety of compounds is being used. There are oxidizing
agents like hydrogen peroxide or chlorine, reducing com-
pounds like sulfur dioxide, heavy metal salts such as
those of mercury and silver, light metal salts like sodium
fluoride or arsenate, acids like boric or salicylic, alkalies
such as slaked lime or trisodium phosphate, alcohols like
ethyl alcohol or phenol, amines, many dyes, and synthetic
or natural drugs. It is not possible that all these so
different compounds react with proteins in the same way.
However, they all lead to the same general result, the
inactivation of bacteria. This becomes understandable
if one considers that proteins are very large molecules
with many different ' ' side chains ' ' so that they offer the
possibility of a great variety of reactions which may have
the same outcome on the whole molecule. Native pro-
teins react very easily and promptly, and it is consis-
tent with chemical experience to assume that reactions
on many different side chains or atomic groups produce
the same physiological result, namely, the denaturation
of the protein, which thereby becomes unfit for further
biological function.

The reason why so many different compounds are be-
ing used is of a practical nature ; it must be sought above
all in the various purposes for which they are employed.
Mercury compounds cannot be used to sterilize food;
salicylic acid will preserve acid fruits, but not neutral
meat; chlorine is an excellent disinfectant for water,
less efiicient in sewage, still less so when organic mat-

62 DISINFECTANTS

ter is concentrated; propionic acid will prevent the
growth of molds, but not of bacteria; sulfa compounds
work well internally, but are poor disinfectants in vitro.

We shall use in this discussion the customary distinc-
tion between disinfectants (called also germicides or bac-
tericides) and antiseptics, though the line of demarca-
tion between them is not sharp. According to the gen-
eral conception, disinfectants sterilize, i.e., they kill all
bacteria present, whereas antiseptics merely prevent
their multiplication. Disinfection is something final,
irrevocable, while antisepsis is something temporary
which can be interrupted or discontinued by the removal
of the antiseptic.

This definition would imply, if taken literally, that
antiseptics never cause death. However, most anti-
septics kill at least a small percentage of cells per hour,
yet death is always very slow, so that even after several
days not all cells are killed. Thus a precise distinction
between disinfectants and antiseptics is possible only
by choosing an arbitrary time limit, and by specifying
various conditions (see below, pp. 63 and 118).

Though these specifications are arbitrary, there is
clearly a fundamental difference between the tempo-
rary, reversible inhibition assumed in the definition of
antiseptics and the permanent, irreversible inhibition
required in that of disinfectants. In some cases, as
with crystal violet, it has been proved experimentally
that the chemical reaction causing reversible inhibition
is different from that which causes irreversible inhibi-
tion, as will be seen below (p. 136).

I. DISINFECTANTS

The difficulty of defining disinfectants has just been
pointed out. To give some precision to the definition
let us assume that we use as a standard method the well-
known technique of the phenol coefficient. Accordingly,
a substance is considered a disinfectant if 5 cc. of the
most concentrated of its commonly used solutions, mixed
with l^ cc. of a 24-hour culture of the test organism,
reduces the number of viable cells so far that, after one
hour, a loojDful of the suspension transferred into broth
will not cause growth.

This definition excludes such compounds as CuSOi,
ZnS04, NaoSOs, all the dyes, sulfanilamide, quinine and
practically all the therapeutic agents. These substances
are, therefore, antiseptics, and will be considered in the
second section of this review. The first section, which
will treat of disinfectants proper, will include alcohol,
phenol, formaldehyde, hydrogen peroxide, the silver and
mercury salts, the mineral acids and the alkalies, the
halogens, a few neutral salts and complex organic com-
pounds. The soaps will also be included in the first sec-
tion, though from purely practical considerations, be-
cause, while soaps really are antiseptics, they are never
used as such.

I. SELECTIVITY OF DISINFECTANTS

It has been realized from the earliest studies of disin-
fection that not all bacteria react alike with the same
disinfectant. Some groups, or some species, are more
sensitive or more tolerant than others. An outstand-
ing example of tolerance is that of mycobacteria to-
wards alkali and chlorine. Another practically impor-
tant example of selectivity is that of the Gram-negative
and Gram-positive species which often react differently
with a given disinfectant. For this reason it has be-

64 DISINFECTANTS

come customary to standardize disinfectants not only
with the Gram-negative Bacterium typhosum, but also
with the Gram-positive Staphylococcus aureus.

The degree of selectivity of various bacteria is not the
same with all disinfectants. Bacterium typhosum and
Staphylococcus aureus are killed by approximately equal
concentrations of acetic acid while, with certain soaps, a
concentration 30 times higher is required to kill the ty-
phoid bacteria than to kill the staphylococci.

Table 13, which has been compiled from the data gath-
ered by McCulloch (1936), gives some conception of the
variations in the degree of selectivity. It shows the dif-
ferences in sensitivity of the two standard test species.
Small differences cannot be considered as due to selec-
tivity since we know that, even within the same species,
different strains may show remarkable variations of tol-
erance. Cases in which one of the test species required
at least 5 times as strong a disinfectant as the other have
been considered indicative of selectivity. Unfortunately,
the published data on selectivit}' show little agreement.
This is due, at least in part, to the inaccuracies in the
determination of the phenol coefficient (see p. 105).

It is important for the theory of disinfection as well
as for the application of disinfectants that most of the
efficient, rapidly killing substances are not very selective.
However, hydrogen peroxide, permanganate, alkalis and
some soaps show a tendency towards selectivity. In the
case of alkalis, for instance, the typhoid bacterium is
killed in 10 minutes at pH 10.02 while the staphylococcus
requires a pH of 12.2, which represents 100 times as many
OH ions.

Besides these differences between the standard test
species, differences occur also in a few other groups of
bacteria in regard to one or several disinfectants. Thus
the resistance of the tubercle group towards alkali and
chlorine is well-known. The waxy membrane of these
bacteria may explain their exceptional behavior. Not so

Dlt^INFECTAXTS

65

QQ DISINFECTANTS

simply explained is the high degree of tolerance toward
iodine exhibited by Pseudomonas pyocyanea which, on
the other hand, is unusually sensitive to acids. Another
remarkable instance of selectivity is that of colloidal sul-
fur which affects the two test species almost equally, but
kills certain plant pathogens at one-tenth or one-thirtieth
of the usual concentration.

With weak disinfectants, such as soap, and with many
antiseptics, the difference in tolerance of the two stand-
ard test species may be hundredfold. Among the out-
standing examples of extreme selectivity are some weak
acids like sulfur dioxide, the dyes, the sulfa drugs, and
many chemotherapeutic agents.

From these observations, certain general conclusions
concerning the mechanism of death can be drawn. If a
disinfectant shows no selectivity, it causes death either
by reacting with a cell constituent common to all spe-
cies, or by causing a reaction which is common to all
proteins and which proceeds at the same rate in all
species. If a substance shows pronounced selectivity, it
reacts on cell constituents which are not alike in the
bacterial species concerned.

II. CONCENTRATION AND DEATH RATE

A classical example of the effect of concentration of
the disinfectant on death is shown in Table 14, which rep-
resents the action of HgCL on the spores of the anthrax
bacillus, and is taken from the first quantitative study
of chemical disinfection by Kronig and Paul in 1897.

The data of this table may be anah^zed on the assump-
tion that the course of death of bacteria is described by
the equation

initial number

Kt=log

survivors

DISINFECTANTS

67

where t is the killing time and K the deathrate constants
This constant was computed for each time interval and
its average for each concentration is given at the bottom
of Table 14.

TABLE 14

Action of various concentrations of mercuric chloride on spores
of B. anthracis. The figures represent the number of viable spores
after various times of exposure. K is the deathrate constant (From
Kronig and Paul, 1897.)

Exposure time
in minutes

M/16 HgCl2

M/32 HgCl2

M/64 HgCl2

2

549

3

323

678*

4

236

5

138

961

6

82

310

7

42

8

19

9

168

10

10

12

1

38

14

0

15

io

178

18

5

20

'41

21

....

3

24

2

25

9

27

i

30

....

0

7

35

....

3

40

....

2

45

1

50

1

K r= 0.22

K = 0.139

K = 0.080

The values of K increase with the concentration, but
not proportionally. In 1908, Watson, using the data by
Chick (1908), showed that the relation of the death rate
to the concentration is best expressed by an equation in
which the concentration c is raised to a power n. This
equation is

initial number

Ktc" = log

survivors

1. See p. 17.

68 DISINFECTANTS

where K is the deathrate constant at the concentration
c = l.

The value of n varies greatly with different disin-
fectants. It may be as low as V2 (which means that the
deathrate constant is proportional to the square root of
the concentration) as Paul and associates (1910) found
for staphylococci exposed to hydrochloric acid. With
many common disinfectants it is near 1, as, e.g., with
formaldehyde, hydrogen peroxide, metal salts, and some
acids. In this case only, the death rate is directly pro-
portional to the concentration. The highest coefficients
recorded (see below. Table 15) were obtained with alcohols
at concentrations of 26 to 40% ; but these high values may
be partly due to osmotic effects or to partial dehydration
of the protoplasm of the cells. Otherwise the highest
values, which are in the neighborhood of 6, were obtained
with phenol and its homologues. This makes the phenol
group quite different from other disinfectants.

We can calculate n from the deathrate constants Ki and
ATo at two different concentrations Ci and C2. The rela-
tion

K,

K,

i-^j

gives

n =

log Ki - log K2

log Ci - log C2

Using the numerical values of Table 14, we have

log 0.22 - log 0.08 ^ Q ^3
log 64 - log 16

The concentration exponent n can also be determined
from the death times, since the death times in experi-
ments with identical inocula are inversely proportional to
the deathrate constants :

DISINFECTANTS 69

Ki t2

K2 ti

We have, therefore

- (-^T-

n =

log ta - log ti

log Ci - log C2

Chick (1908) found the following death times for wash-
ed cells of Bacterium paratyphosum exposed to silver ni-
trate :

Concentration : Death Time :

0.085% 0.75 minutes

0.017% 1.5

0.0085% 2.5

0.0017% 6.5

0.00085% 22.5

0.00017% 56

0.000085% 140
The exponent can be computed from any two sets of data.
The two extremes give

^ log 140 - log 0.75 ^^rjn

log 1000 * '

The next two values give

log 56 -log 1.5 _ ^ r,Q
""= — b^lOO ^'^'

The significance of the concentration exponents will
appear more clearly in a specific example. At a con-
centration of 0.7%, formaldehyde and phenol have about
the same disinfecting power and kill Bact. coli in about
100 minutes. If the concentration is doubled, formal-
dehyde, for which n = 1, kills in 50 minutes, while phe-
nol, with n = 6, will kill in 100/2<^ = 100/64 = 1.6 min-
utes. If the disinfectants are diluted with an equal vol-
ume of water, 0.35% formaldehyde will kill in 200 min-
utes while phenol requires 100 X 2'' = 6400 minutes =
4.5 days.

70 DLSIAFECTANT.S

This can be shown graphically by plotting the loga-
rithms of the death times against the logarithms of the
concentration. This type of plotting must result in a
straight line, as the following reasoning indicates. In the
determination of death times, the right side of the
equation

initial number
Ktc" = log -

survivors

is not affected by changes in concentration because all
tests of a series begin with the same initial number and
end with the same number of survivors, theoretically
very few, practically none. K, the deathrate constant
for c = i, is also constant. We therefore obtain

tC^ = M

where M is a constant. Dividing the two members by t
and taking their logarithms, we have

n log c = log M - log t

where log M is a constant. Therefore log t is a recti-
linear function of log c, and n controls the slope of the
line. Figure 16 shows the different slopes produced by
different values of n. Figure 17 shows the curves for
phenol and formaldehyde crossing each other. Curves
for other disinfectants may be found in Figure 22.

The meaning of the concentration exponent in the le-
thal reaction of bacteria is not clear. In chemical reac-
tions, it frequently indicates the number of reacting
molecules. In the case of disinfection, this seemed rather
improbable to Watson (1908) because n is not always an
integer. Further, the not uncommon value n = 0.5
should indicate that two molecules of bacterial protein
react with one molecule of disinfectant, which seems
highly improbable for such disinfectants as HCl or 0:-.
But the assumption that n corresponds to a given num-

DISINFECTANTS

71

10 20 30 40 50 MINUTES 80 90 00

©V

LOG. DEATH TIME VS. LOG. CONCENTRATION

2 3 5 MIN. 10

Fig. 16. Two methods of representing the relation between concen-
tration of disinfectant and death time. The different curves illustrate
how the "concentration exponents" n affect the results.

ber of reacting molecules becomes definitely untenable
if the observation by Tilley (1939) can be confirmed,
that the exponent decreases when the temperature in-
creases. Though Tilley 's data show a good deal of
fluctuation, they clearly indicate this influence of tem-
perature (see, in particular, the figures for the concen-
tration exponents of phenol and ethyl alcohol in Table 15).

72

DISINFECTANTS

The alternative explanation is that the disinfectant
acts as a catalyst, speeding up the very slow denatura-
tion of some essential protein. Native proteins are never
stabile. With such catalytic action, the concentration
exponent need not be an integer, and it is usually greatly
affected by temperature. However, the determination of
n m disinfection is too inaccurate to permit decision be-
tween the two theories.

3 %

\

N

%

V

^

\

\

s

\

^

^

^

Nol

\

\

1

— '

\

s.

"^

■~-^

-^

'■ — 1

\

? '

\.

i \

0

0

toth
? 1

1
1

Tim«

7

f) ^

Q^

_i

f) ?

\

£11

^hrv

15

L2'

1.0.
0.9
08

as

Fig. 17. Effect of various concentrations of formaldehyde and of
phenol on the death time of Bacterium coli. Abscissa: death time, on
logarithmic scale; ordinate: per cent concentration, on logarithmic
scale. (Original data.)

Concentration exponents have been measured occasion-
ally since Chick (1908) and Watson (1908) first called at-
tention to them. Only two extensive studies with large
numbers of experiments have been published, those of
Paul, Birstein and Reuss (1910) on different acids, and
those of Tilley (1939) on the phenol group. Paul and
associates used the method of death rates, Tilley the
method of death times. Some data by these and other
authors are shown in Table 15.

Not only do the values of n fluctuate considerably in
different experiments, but Tilley found, with Bad.
typhosum and phenol, that they fluctuated in the same

DISINFECTANTS

73

experiments as follows : 9.35, 5.59, 11.17, 8.76 ; average :
8.96. A second identical experiment gave n = 7.04 as
the average of four determinations. If such fluctua-
tions are obtained by experienced investigators, it is
evident that concentration exponents are not reliable
enough to serve as a basis for speculations on the mech-
anism of the lethal reaction.

TABLE 15

Concentration exponents of some disinfectants.
J. Acids (Paul, Birstein and Reuss, 1910)

Number of
experiments ,

Average exponent based on
concentration of

Acid

H ions

Hydrochloric acid
Acetic acid
Butyric acid

4
3
3

0.52
0.98
2.04

0.52
1.95
4.02

77. Alcohols (Tilley, 1939)

Bact. typhosum

Staph, aureus

10°

20°

30°

40°

10°

20°

30°

40°

Phenol
o-Cresol
p-Cresol
o-Butyl phenol
p-Butyl phenol
Resorcinol
Ethyl alcohol
n-Butyl alcohol

7.9
8.3
8.9

4"'6
12.7

7.5
7.9
8.4
9.2
9.2
5.1
11.4
11.9

5.9
5.5
6.4
7.2
9.3
5.2
8.8
10.3

5.2
5.1
5.3

6.5

7.8
8.2

4'4
11.4

6.5
7.6
8.7
8.5
8.9
5.0
11.1
11.8

6.4
8.1
9.4
9.1
8.1
6.2
8.5
10.1

5.4
G.8

7.0

777. Various Compounds

Phenol 5.5 (Chick, 1908)

Formaldehyde 1.0 (Gegenbauer, 1922)

Hydrogen peroxide 0.5 (Reichel, 1908)

Hydrogen peroxide 1.7-2.6 (Rahn, unpublished)

Silver nitrate 0.9 (Chick, 1908)

74 DISINFECTANTS

There is still more inconsistency with disinfectants
which combine readily with various kinds of organic
matter. A typical example is that of the concentration
exponent of mercuric chloride which was found to have
the following values in various cases :

0.25 Alfred Miiller, 1920, with anthrax spores;
0.5 Gegenbauer, 1921, with Staph, aureus;

3.8 Chick, 1908, with Bact. typhosum;

4.9 Chick, 1908, with anthrax spores.

It seems c( rtain that, with this compound, different
amounts of organic matter in the test solution are re-
sponsible for different values of n. If in one case the
proteins of the medium precipitate 90% of the HgCL
present, while in another they precipitate 10%, the values
of the concentration coefficient must necessarily be very
different.

Another unsettled point is the degree of specificity of
the organisms. Different bacteria might give different
concentration exponents with the same disinfectant. The
only systematic research known to the author on this
subject is that cited in the middle section of Table 15.
There are differences between the exponents obtained with
Bact. typhosum and Staph, aureus, but they are not con-
sistent.

III. TEMPERATURE AND DISINFECTION

The preceding chapter bears out the general assump-
tion that disinfection is the result of a chemical reaction
of measurable velocity. Such reactions are usually af-
fected by changes in temperature. The rate of most
chemical reactions increases when the temperature rises,
and so does the death rate. The increase in rate differs
with, the type of chemical reaction, and a quantitative de-
termination of the effect of temperature on disinfection
is therefore important in the search for the "lethal

DISINFECTANTS 75

reaction". A temperature coefficient of disinfection sim-
ilar to that of chemical reactions excludes any theory
of disinfection based on rates of diffusion, e.g., that by
Henderson Smith, 1921^ because the temperature etfect
on diffusion is much smaller than that on most chemical
reactions.

Most biologists are probably more familiar with the
conception of temperature coefficients than with that of
temperature increments, and therefore temperature co-
efficients will be used throughout this discussion, though
the increments would be preferable since they remain
constant while the temperature coefficients decrease
slightly with increasing temperature.

Temperature coefficients of disinfection have been de-
termined by Madsen and Nyman (1907), by Chick (1908
and 1910), by Paul, Birstein and Reuss (1910), and by
many others. Table 16 lists some of the more recently
determined values of the temperature coefficient when
disinfection was accomplished by various means (by al-
cohols, Tilley, 1942; by chlorine, Charlton and Levine,
1937; and by high acidity, Cohen, 1922).

A simple relation exists between temperature coeffi-
cients and death times. If the deathrate constant in-
creases Qi times for every degree increase in tempera-
ture, and if K is the constant at the temperature T, then
the constant at the temperature T -i- N is KQi^ and the
formula for death at this temperature is

KtQ,^ = log initial number
survivors
In any set of experiments to measure death times, the
values for K, initial number and survivors are the same,
and we have the simple relation

tQi^ = constant = A
and therefore, N log Qi = log A - log t.
This indicates that a straight line will be obtained if the

1. See p. 35.

76

DISINFECTANTS

TABLE 16

Temperature coefficients of disinfection
I. Disinfection 'by various alcohols. (From Tilley, 1942)

Disinfectant

Bad. typhosum

Staph.

aureus

10-20°

20-30°

30-40°

10-20°

20-30°

30-40°

40-50°

Phenol

5.8

8.3

8.4

5.1

3.9

4.0

7.4

Ortho-cresol

6.6

5.1

6.9

5.4

4.2

4.3

10.0

Para-cresol

6.1

5.8

5.6

4.3

4.1

4.4

9.0

Resorcinol

7.1

7.1

8.8

3.1

3.2

4.4

....

Ethyl alcohol

43.0

54.0

13.0

9.0

....

n-Butyl alcohol

31.0

40.0

11.0

9.0

....

II. Disinfection by chlorine at various acidities. (From Charlton
and Levine, 1937)

PH

Bacillus metiens

25-36°

35-45°

45-55°

6.0

8.7

6.1
3.7

5.4
3.1

5.4
3.5

///. Disinfection by high acidity, at pH 3.5. (From Cohen, 1922)

Bad. coli

Bact. typhosum

0-10°

10-20°

20-30°

0-10°

10-20°

20-30°

2.1

4.4

3.8

1.6

1.5

1.8

logarithms of the death times are plotted against tem-
perature. Figure 18 shows a few of Chick's data plotted
in this way. The curves representing the different phenol
concentrations are nearly parallel, indicating the same
Qio, whereas the curve for AgNOa has a different slope,
which means that the reaction has a different Qio.

For practical purposes, it is customary to determine
which concentrations will . produce the same degree of
sterilization at different temperatures. McCulloch (1936)
found that Bact. typhosum will be killed in 10 minutes
at 2° by 2.0% phenol, at 20° by 1.18% and at 40° by 0.56%.

DISINFECTANTS

77

From such data the temperature coefficient can be de-
termined if the concentration exponent is known. The
equation is

KtiCi" = Kt2C2"QN

where ^n is the temperature coefficient for the interval
A^ in degrees centigrade. For equal disinfection times,
we have

«-=(^j

In the above example, assuming n = 6, we find from 20
to 40°,

Q,o = (1.18/0.56)« or Q,o = (1.18/0.56)^ = 9.3.

For the interval from 2 to 20°,

Qi8 = (2/1.18)'' or Qxo = 1.695^^^« = 5.8.

When n = l, the ratio of concentrations equally efficient
at different temperatures equals the temperature coeffi-

Fig. 18. Effect of temperature on the death time of Bacterium para-
typliosuni treated with phenol. Abscissa: death time, on logarithmic
scale; ordinate: temperature. (From data of Chick, 1908.)

78 DISINFECTANTS

cient. McCuUoch found for formaldehyde {n = 1) the
following concentrations equally efficient: At 2°, 5%; at
20°, 2.5% ; and at 40% 0.5%. This gives Q^s = 2 and
Q20 = 5, and consequently Qio = 1.5 and 2.2 respectively
for the two ditferent intervals.

The large majority of experiments on the role of tem-
perature in disinfection indicate that the effect is similar
to that on chemical reactions. The temperature coeffi-
cients average about 2. Often they are very much high-
er. Rarely are they consistently less than 2, and if so,
this also can usually be explained on the basis of a chem-
ical phenomenon, as in the case of alkali whose action
is not affected much by temperature.

The general trend of results is thus in conformity with
the explanation that disinfection is due to a chemi-
cal reaction of the disinfectant with some cell con-
stituents. It excludes the explanation of the logarithmic
order by a graded rate of diffusion because a tempera-
ture increase of 10 degrees would certainly not be suf-
ficient to double the rate of diffusion. But some observa-
tions have been recorded which remain unexplained, as
e.g., the higher temperature coefficient of very young
cells as compared with old ones (Chick, 1910). A careful
study in the light of the latest knowledge of temperature
effects in biology seems necessary to clarify the role of
temperature disinfection.

TEMPERATURE COEFFICIENTS OF LIFE AND DEATH

The above-mentioned experiments on the influence of
temperature on death processes contain a fact of impor-
tance for the interpretation of temperature effects on
life processes. Most organisms cease to grow when their
temperature is lowered to a few degrees above the freez-
ing point of water or cell sap. This cessation of growth
is not abrupt. If growth ceases at 5°, it is still exceed-
ingly slow at 10° and not very rapid at 15°. The tem-
perature coefficient between 0 and 5° is infinite, between

DISINFECTANTS 79

5 and 10° it may be 5 to 8, between 10 and 15°, about 3
to 4, and it thus gradually approaches the value of 2 to 3
which we find near the optimal temperature. The energy-
furnishing processes behave similarly. Although en-
zymes continue to function below the minimal tempera-
ture of growth, down to the freezing point (Foter and
Rahn, 1936; Rahn and Bigwood, 1939), their temperature
coefficients increase greatly near the freezing point.

Quite contrary to growth rate, death rate displays a
fairly uniform temperature coefficient as far down as
the freezing point. While the number of experiments on
record on that subject is not great\ there is no instance
showing an increase similar to that of the life functions.
Death proceeds more slowly at lower temperatures, but
the decrease of the rate is very slight, its Qio is about the
same as at 20 or 30°, and not several times as large,
as we find it with growth and fermentation. The state
of dormancy induced by low temperature does not in-
terfere with death by chemical poisons.

Since death is a destruction of the mechanism of
growth or cell division, the constancy of the temperature
coefficient indicates that the essential molecules involved
in this mechanism are not impeded much by low tem-
peratures in their reactivity with poisons, while they
are impeded in their reactivity with metabolites. This
is important for the explanation of the great retarda-
tion or complete inhibition of growth and energy pro-
duction of the cell.

One of the several theories offered to account for this
phenomenon is that of an increase in viscosity of the
protoplasm. According to Belehradek (1935, p. 160),
''such an increase would considerably hinder free move-
ments of reacting molecules with the result that the
biochemical reactions in the cell would be brought to a
standstill." But the hindering of free movement of re-

1. See Table 16.

80 DISINFECTANTS

acting molecules should also apply to those of the dis-
infectants. As the latter are evidently not hindered at
all, and as the increased viscosity of protoplasm does
not influence appreciably the death rate, it does not seem
likely that it should influence the growth rate and there-
fore, it is probably not the cause of the cessation of life
functions at low temperatures.

IV. INTERFERENCE OF FOREIGN MATTER WITH DISINFECTION

Sterilization of water supplies and sterilization of in-
struments are perhaps the only cases where bacteria are
exposed to disinfectants in the absence of organic matter.
In most cases of disinfection, proteins are present, and
frequently other organic or inorganic substances. It is
important, therefore, to know to what extent foreign
matter interferes with sterilization.

It has been stated before that probably all disinfec-
tants produce their killing effects by reacting with some
protein in the cell. They must be expected to react also
with proteins outside the cell. Obviously, disinfectants
will be weakened, or their action may even be com-
pletely nullified if they produce insoluble compounds
with foreign materials. Such is the case for silver ni-
trate in presence of chlorides, for mercury salts in pres-
ence of hydrogen sulfide, and for hydrogen peroxide in
open wounds, where it is decomposed by catalase.

A particularly important instance is that of the disin-
fection of the living body in chemotherapy where the
compounds of the blood and of the tissues may tend to
combine with the toxic agent, thereby decreasing its con-
centration and efficiency.

Different disinfectants will be affected differently by
the same foreign substance, and this is one of the
reasons why so many different kinds of disinfectants are
being used.

DISINFECTANTS 81

Chick and Martin (1908) made some experiments on
the interference of proteins with disinfection. They
found that phenol was not much affected by 10% blood
serum. In the average of 4 experiments, the concen-
tration of phenol had to be increased only from 0.806 to
0.924% to accomplish sterilization in 15 minutes. Ex-
pressed in terms of death time, this would mean
15(0.924/0.806)^ = 34 minutes, an increase of 125%,
which is relatively little. Mercuric chloride (0.05%) was
considerably more weakened by blood serum, as shown
in the following table :

Cone, of blood serum (per cent) : 0 5 10 20 30

Death time (in minutes) : 7.2 10 14.2 39 62

Per cent increase in death time : 0 39 97 340 760
Animal charcoal interfered especially with emulsified dis-
infectants. Dried feces acted partly chemically, partly
by adsorption.

Many disinfectants have been tried in the presence
of organic matter, serum, blood, milk, urine, feces, etc.,
but usually only the phenol coefficients are recorded and
this gives very little information on interference by
foreign matter because the phenol coefficients include the
effect of organic matter upon phenol as well as upon the
disinfectant under test.

Our own data with skimmed milk diluted 1:10 show
for phenol, formaldehyde and hydrogen peroxide a pro-
longation of the death times by about 20 to 35% while,
with mercuric chloride, the prolongation was about
1000% and with mercurochrome, 10,000%. Washed bac-
teria were used in these experiments in order to exclude
all foreign matter from the controls. With the customary
method of placing 0.5 cc. of the bacterial culture into 5
cc. of disinfectant, a considerable amount of organic
matter is introduced, namely, 1500 ppm. of peptone and
meat extract. This is sufficient to interfere with strong
disinfectants capable of killing bacteria in very low con-
centrations. The difference of procedure explains the

82 DISIXFECTAXTS

great difference between the results of our own experi-
ments and those by Chick and Martin, with HgCl2.
These authors measured the weakening effect of blood
serum not by comparison with water, but by comparison
with a solution of about 0.15% meat extract plus pep-
tone, which eliminated a good portion of the mercury
ions, while in our controls no protein was present except
that of bacterial cells.

The efficiency of chlorine and iodine is greatly de-
creased by very small amounts of organic matter. For
city water supplies, 0.2 to 0.5 ppm. of chlorine is suffi-
cient to bring about rapid sterilization while 7 to 10
ppm. is the amount ordinarily used for sewage, which
contains only 600 ppm. of foreign matter, less than is
present in the medium used for the determination of the
phenol coefficient.

The H and OH ions are very important foreign sub-
stances. Some disinfectants and antiseptics act very
differently at different pH values. In general, however,
with strong disinfectants, the effect of a change in pH
is not very great. There is only one strong disinfectant
which shows a very great response to changes in acidity,
namely chlorine. Some antiseptics are made absolutely
harmless by a change in pH. This is especially true with
such compounds as benzoic, salicylic, and sulfurous
acids, which are efficient only at low pH, and also with
the dyes, all of which will be discussed in detail later.

V. INTERFERENCE OF ANTIDOTES WITH DISINFECTION

Antidotes are not employed very often in man's strug-
gle with bacteria. When chemical compounds are used
to kill bacteria, either they can be removed by simple
rinsing (skin, instruments), or the sterilized materials
are discarded afterwards (excreta, dead animals). For
materials to be used after sterilization disinfection by

DISINFECTANTS 83

heat has many advantages (e.g., in foods). However,
there are cases in which heat cannot be used, as in the
sterilization of certain foods, or of catgut, and in chemo-
therapy.

An example of the use of disinfectants in foods and of
the subsequent removal by an antidote is the experimental
sterilization of milk for cheese making by hydrogen per-
oxide which is later destroyed by catalase. This method
has been used by Curran, Evans and Leviton (1940) in
the study of bacteriology of cheese ripening. Another
example is the use of SOo in the fermentation industries.
The antidote in this case is atmospheric oxygen which
very slowly oxidizes the SO. to sulfuric acid which at
these great dilutions is harmless.

In chemotherapy the entire body may act as an anti-
dote. Concerning the role of the body, however, a dis-
tinction should be made between two very ditferent modes
of action, namely, a chemical combination between the
disinfectant and some components of the bodj^, and a
mere dilution caused by partial elimination of the poi-
son. Only in the first case can we really speak of the
body reacting as an antidote.

The difference between these two modes of action is
well illustrated by the experiments of Gegenbauer which
will be discussed a few pages below. Staphylococci
treated with HgCL could be made to multiply again
by repeated washing with water if the exposure had
not been longer than one hour. After long exposures,
washing did not result in recovery, but an antidote pro-
ducing an insoluble compound was capable of making the
bacteria multiply again.

Washing or diluting cannot repair injury. Such op-
erations can only remove conditions which prevent the
functioning of the uninjured cell mechanisms; they re-
move, so to speak, the ''monkey Avrench" from the en-
gine. This is illustrated by the fact that most bacteria,

84 DISINFECTANTS

though they cannot multiply in acid media, such as cider,
are however not killed, and start to multiply as soon as
the acid is neutralized. (The production of this type of
''suspended animation" is usually called antisepsis.) The
antidote does more than washing or diluting, it repairs
the primary injury done by the disinfectant to the cell
mechanism, but not the further injury resulting from the
abnormal or interrupted functioning of the injured mech-
anism. When this secondary injury has advanced so
far as to be irreparable, the cells cannot be revived by
the antidote.

The application of disinfectants to wounds must be
considered from this viewpoint. It has been shown that
bacteria, affected chemically to the point where they
cannot multiply in broth, may still cause infection when
injected in animals. The rapid circulation of liquids in
the body results in a washing of the bacteria. Besides,
certain compounds of the blood or tissue juices may
diffuse into the bacterial cell and react chemically with
the disinfectant therein.

VI. PROPERTIES OF VARIOUS GROUPS OF DISINFECTANTS

It is not our purpose in this discussion to describe the
effect of each bactericide, or the advantages and disad-
vantages of its use in various applications. Such de-
scriptions will be found in medical handbooks (see, for
example, "Disinfection and Sterilisation" by McCul-
loch). We are concerned here mainly with the nature
of the lethal reaction.

1. Acids. Strong mineral acids act essentially by
their hydrogen ions. Kronig and Paul demonstrated this
in their classical study on the physical chemistry of dis-
infection (1897), in which they found that most of the
strong acids kill bacteria at approximately the same
rate when diluted to the same normality. (Since many

DISINFECTANTS 85

bacteria will die at pH 2, which means 10 mg. H ions per
liter, or 10 ppm., hydrogen ions must be considered very
efficient disinfectants.)

In a few cases, however, such as that of HNO3 and HF,
not only the H ions but the anions also act on proteins ;
the death rate then is higher than the pH would indi-
cate. With weak organic acids such as benzoic and sali-
cylic, the undissociated molecule itself is toxic. The same
is true for SO, (see p. 159). The strongly dissociated
CCI3COOH acts by its anions as well as its cations.

The effect of pH on the colloidal state of protein is
well known, and denaturation by a high concentration
of hydrogen ions is to be expected, provided that the
acidity of the medium can affect the pH of the cell con-
tents. But different proteins have different isoelectric
points, and a wide range of different responses to pH
changes is not surprising. In fact some yeasts as well
as some Aspergilli can multiply at pH 2.0 ; certain sulfur-
oxidizing soil bacteria are known to exist even at pH 1
(Starkey and Waksman, 1943), while some urea-splitting
bacteria grow best at pH 8.5 to 9.0.

2. Alkalies. Alkalies are commonly used as bac-
tericides especially in soaps and washing powders. It
has been shown that some compounds like trisodium
phosphate, sodium metasilicate, and slaked lime act
largely in proportion to their hydroxyl ions (Myers,
1928 and 1929, Tilley and Schaffer, 1931). But the tol-
erance for these ions varies greatly with the species. The
lethal action is usually explained by the prompt denatur-
ation of proteins by excessive alkali.

3. Heavy Metal Compounds. Kronig
and Paul (1897) proved that heavy metal salts act pri-
marily by their metallic ions. These authors decreased
the ionization of HgCL by the addition of NaCl, and
showed that, as the concentration of the latter was in-
creased, the efficiency of the mercury salt decreased.

86 DISINFECTANTS

An extensive study of the mechanism of mercuric
chloride disinfection was made by Gegenbauer (1921).
He found that HgCL is not only adsorbed on coagulated
serum proteins, but that it is also dissolved in them, and
that it can be washed out again. Besides, an insoluble
mercury proteinate is formed which cannot be washed
out, but the normal protein can be restored upon the
addition of H2S which forms HgS. Gegenbauer then
applied these findings to Staphylococci. When the cells
had remained for an hour in a solution which contained as
much as 2% HgCL, they would grow if repeatedly washed
with water. But after 2 hours of exposure they could
not usually be revived by washing. However, when the
antidote HgS was applied, bacteria from a 2% HgCU so-
lution could be brought back to life after 6 hours, but not
after 11 hours. With 0.1% HgCL, the cells became viable
upon HoS treatment even after 36 hours. The results are
shown in Figure 19.

According to Gegenbauer, the actual poisonmg pro-
ceeds in three stages: first, the bichloride of mercury
dissolves fairly rapidly in the protoplasm of the Staphy-
lococci and probably interrupts all cell functions. This
proceeds according to a definite partition coefficient.
Then a slower process begins, the formation of a mer-
cury proteinate. While the first process can be reversed
by washing, the proteinate cannot be removed by wash-
ing, and after about 2 hours, its formation has gone so
far that in every cell some vital protein has been changed.
This process, however, becomes reversible on the addi-
tion of H2S, though only within a limited time. There-
fore, we must assume a third reaction, namely the de-
struction of some essential life function. Either some
catabolic process is accelerated by mercury, or the mer-
cury proteinate itself may undergo a further change
which is irreversible. The rate of this third process is
approximately proportional to the square root of the
mercury concentration.

DISINFECTANTS 87

3.0%

to

0.029

\

\
\

\

\

\

X

1

%

\

>

I

I

L

\

\

\

\

2

\
3

4

V-~^ 10 20

30 40 50

79 100

T-

\
\
\

Fig. 19. Relation between the concentration of HgCK'. and the time
during which staphylococci can be exposed to that disinfectant and
still recover. The curve at the left indicates the limits of recovery after
washing, the curve at the right, the limits of recovery after use of H2S
as antidote. Abscissa: time in hours, on logarithmic scale; ordinate:
percent concentration on logarithmic scale. (From data of Gegenbauer,
1921.)

Many organic mercury compounds have been tested in
the hope of finding disinfectants which are less toxic and
less irritant to mammals than the bichloride. Some of
these newer disinfectants are in fact more powerful than
the bichloride against bacteria and relatively less toxic
to the hosts. Their use, however, is not as general as
that of the inorganic compounds, because the properties
of the organic radicals affect the chemical behavior of
the compounds. This is exemplified in the great de-
pendence of the effect of mercurochrome upon the acidity
of the substrate (a phenomenon which is typical of all
dyes and therefore is not surprising in mercurochrome).

Other organic mercury compounds are highly selec-
tive, like metaphen, merthiolate and phenyl-mercuri-ni-
trate. These three compounds are stronger disinfectants
than the bichloride of mercury — mercurochrome is not.

The organic mercury compounds have no mercury ions,

88 DISINFECTANTS

and no suggestions have been made concerning their mode
of action.

The only other heavy metal compounds used extensive-
ly as disinfectants are those of silver. The silver salts,
nitrate, lactate, and citrate, act probably through their
metallic ion, as in the case of the mercury salts. Their
instantaneous reaction with chlorides to form insoluble
AgCl limits their usefulness greatly.

The preparations of silver metal in colloidal form, with
trade names like argyn, argyrol, solargentum, silvol, etc.,
are of particular interest. Metallic silver does not react
with chlorides or proteins as silver ions do, but it is only
a very weak disinfectant. According to Leonard (1931),
a suspension containing as much as 1 gram of colloidal
silver in 100 cc. does not kill the test organisms Bact.
coll and Staph, aureus, though suspensions of 1 gram in
10,000 are sufficient to prevent multiplication.

This leads us to the problem of the oligodynamic action
of metals which is still much debated. The first obser-
vations were made with metallic copper whose disin-
fecting action has been proved by several investigators.
The lethal effect cannot be due to copper ions going
into solution from the metal because the solubility of
metals is immeasurably small, and a corresponding
amount of copper ions from any copper salt has not the
least inhibiting effect. Even a solution of 5% copper
chloride does not kill Staphylococcus aureus in one hour.

If direct contact of the cells with the metal were nec-
essary to bring about death, the latter might be explained
by a chemical combination of some essential cell con-
stituent with the metal, or by a physical action, such as
interference with some electric potential in the cell.
However, no conclusive proof has been given that direct
contact is necessary.

Most experimenters agree that oxygen is necessary
to bring about the lethal effect of metallic copper sur-

DISINFECTANTS 89

faces. This suggests a chemical rather than a physical
effect, perhaps the formation of a peroxide. But the
amounts of hydrogen peroxide formed on metal sur-
faces in water by oxygen are far too small to explain
the oligodynamic action. However, a combined action
of peroxide and metal ions may be sutficient to kill
bacteria. Dittmar, Baldwin and Miller (1930) had ob-
served that the lethal action of hydrogen peroxide is
speeded up nearly 100 times by the presence of 0.1 mil-
limoles of Cu++ and Fe+++ ions. With 6 mg. Cu++ per
liter. Staph, aureus was killed in 10 minutes by 0.025%
H2O2, and in 30 minutes by 0.008% H,0..

The use of silver surfaces has been proposed to ster-
ilize water and even milk. Bechliold (1918) coated
particles of coal and sand with a thin layer of silver and
used them for the sterilization of water. Krause and
Bergmann (1928) produced the so-called ''katadyn" by
turning molten silver into a fine foam. The crushed
solidified foam bubbles are claimed to disinfect water
upon contact. The results obtained with milk, however,
are not very encouraging.

Colloidal silver differs from these silver preparations
only in having a much larger surface. No records could
be found which would throw light on the important
question whether oxygen is necessary for the disinfec-
tion by colloidal silver preparations.

4. The Alcohols. Of the aliphatic alcohols,
only ethyl alcohol is used as a disinfectant, and it is
rather weak. Its strongest action is at a concentration
of 70%. Absolute alcohol is less effective because it is
dehydrating, and proteins without moisture are not
easily denatured.

The rate of denaturation of different native proteins
by ethyl alcohol varies greatly. Some enzymes lose
their activity in alcoholic solutions rather slowly.

A surprisingly high concentration exponent was
found by Tilley for ethyl alcohol (Table 15).

90 DISINFECTANTS

Of the cyclic alcohols, phenol is the best known, but
the cresols, thymol, resorcinol, etc., are even stronger
disinfectants than phenol. One of the early attempts
at explaining the action of phenol is that of Reichel in
1909. He showed that phenol dissolves in coagulated
serum albumin, and that the fraction of phenol going
into the albumin can be increased by the addition of
salt. Salt also increases the disinfecting power of
phenol, and the death times with various NaCl con-
centrations increase in proportion to the partition co-
efficient of phenol between water plus salt, and albumin.

This not only explains the effect of the salt, but proves
that the lethal action is brougli.t about only by the pro-
tein-dissolved fraction of the jDhenol. The lethal re-
action itself remains unknown. Denaturation would be
the simplest assumption, but we know that some pro-
teins are not readily denatured by the higher alcohols.
Phenol and thymol are frequently used in experiments
with enzymes to prevent bacterial action; the denatura-
tion of enzymes (which are proteins) by these com-
pounds must therefore be very slow. Of course, this
refers only to dilute solutions, and their rate of disin-
fection (and perhaps of denaturation generally) in-
creases as the 6th or 7th power of the concentration. A
doubling of the concentration would make denaturation
2*^ = 64 times as rapid. As stated above (p. 72) the
high concentration exponent and its change with tem-
perature make it appear possible that phenols are mere-
ly catalysts speeding up some natural deterioration
which is normally so slow as to be balanced by the cell's
repair mechanism.

5. Formaldehyde. Formaldehyde is the only
aldehyde known to have a strong disinfectant power,
though not for all organisms ; molds are not very sensi-
tive to it. It has one outstanding property, namely,
it retards multiplication in very great dilution (see p.
129).

DISINFECTANTS 91

Formaldehyde reacts very promptly with the amino
groups of proteins. (This reaction is used in the quan-
titative determination of amino groups.) Naturally,
such proteins are then incapable of any physiological
function.

6. Oxidizing Agents. The two best known
oxidizing compounds are hydrogen peroxide and per-
manganate. They react very differently, however. Per-
oxide does not attack proteins or other organic matter
readily, but in tissues it is destroyed by the catalase. It is
a somewhat more efficient disinfectant than formalde-
hyde, and resembles the latter by acting proportion-
ately to its concentration, the concentration exponent
being approximately 1. Permanganate reacts prompt-
ly with organic matter, changing to the inert brown
MnOa and thereby losing its efficiency. In pure water,
it is more efficient than peroxide, but its usefulness is
greatly limited whenever organic matter is present.
The concentration exponent of permanganate is approx-
imately 4, indicating a rapid loss of power by dilution.

The lethal reaction with hydrogen peroxide is prob-
ably an irreversible oxidation of some vital protein.
This oxidation is not due to nascent oxygen liberated by
the catalase in the bacterial cells, for the streptococci
and lactobacilli which do not contain catalase are nev-
ertheless killed by peroxide.

The rate of oxidation of various chemical substances
by hydrogen peroxide is greatly accelerated by the si-
multaneous presence of small amounts of Fe+++ and
Cu++ ions, and Dittmar, Baldwin and Miller (1930)
found this to be true also for disinfection. With 0.1
millimol each of Fe2( 804)3 and CuSOi, the phenol co-
efficient of hydrogen peroxide increased with Bad. coli
from 0.014 to 1.4, and with Staph, aureus from 0.012
to 1.2. A combination of potassium dichromate with
cobaltous or manganous sulfate produced the same re-
sult.

92

DISINFECTANTS

As to permanganate and chromate, their reaction with
organic compounds is so ubiquitous that it is impossible
to point to any specific reaction as the cause of death.

7. The Halogens. Chlorine and iodine re-
semble permanganate in their very rapid reaction with
many different organic compounds, but they are far more
powerful disinfectants than permanganate. They are
used widely : chlorine in the disinfection of water supplies
(which contain practically no organic matter to weaken
its action) and iodine externally, in medical practice,
mostly as a skin disinfectant. Their reaction with organic
matter is so great that the phenol coefficient, which is
about 200 when determined in the ordinary way, with %
ml. of culture, rises to 2,000 when determined with wash-
ed bacteria (see p. 110).

Chlorine disinfection has several unusual features. One
is its great sensitivity to pH. The data by Tilley and
Chapin, 1930 (Table 17) will illustrate this. According to
Holwerda (1928) the sensitivity to pH can be explained
by the reaction of chlorine with water:

CI2 + H,0 — > HOCl + HCl
and

HOCl — > H+ + OCl-

TABLE 17

Death

times of spores of B. anthracis treated Avith chlorine

in

different

buffer solutions. (From Tilley and Chapin, 1930.)

pH

Available

chlorine

ppm.

Some spores
alive after

All spores
dead after

10

100

120 min.

....

9

30

60 "

120 min.

9

40

45 "

60 "

9

50

15 "

30 "

8

10

15 "

30 "

7

10

15 "

30 "

6

3

120 "

6

4

15 "

5

3

15 "

30 "

4

2

30 "

45 "

3

2

30 "

45 "

DISINFECT AN Ti^ 93

Two dissociation constants of HOCl are given in
chemical literature which ditfer greatly, namely 3.7X10-^
and 6.7X10-^^ The latter is supposed to be the more ac-
curate ; it certainly comes from a later measurement, but
Holwerda based his calculations on the first, older con-
stant. His experimental data agree fairly well with his
assumption that the undissociated HOCl is the toxic part
of chlorine dissolved in water. But his experiments were
largely limited to alkaline and neutral solutions. The
increased efficiency of chlorine by a decrease in pH from
7 to 3 (Table 17) cannot be explained in this way. The
newer dissociation constant does not give as good a cor-
relation as the older one.

Another unusual feature of chlorine disinfection is the
non-logarithmic type of survivor curve, as shown pre-
viously. Two very extensive investigations, one by Charl-
ton and Levine (1937) and the other by Mallmann and
Ardry (1940), bear out this observation. The explan-
ation must be sought in the usual mode of action of
chlorine as illustrated in the following case reported by
Anderson and Mallmann, in their recent study of the
penetrative powers of disinfectants (1943). They ob-
served under the microscope the effect of various disin-
fectants upon freshly hatched long-tailed strongylid lar-
vae (parasites of horses). Under the action of chlorine
''the thin layer of cuticle which forms the tail of the or-
ganism dissolved first, followed by a splitting or shred-
ding of the posterior end of the organism, simultaneously
with the appearance of bubble-like rupture of the cuticle
along the whole length of the organism." The action of
phenol was quite different. At a concentration of 1 :100 it
"produced death of the larvae in 45 minutes without any
external changes of the organisms." In their summary,
they state: "Chlorine acts only by oxidation, i.e. first
destroying the cell surface."

These observations and conclusions speak unequivo-
cally in favor of the explanation of chlorine action pre-

94

DISINFECTANTS

viously given by the present writer on the basis of theo-
retical considerations. A fairly large number of mole-
cules of the cell surface must be destroyed to produce
an injury from which the cell cannot recover. The
theoretical survivor curve when death results from the
inactivation of many molecules agrees in every respect
with that observed in chlorine disinfection.

No definite value of the concentration exponent of
chlorine can be obtained because of the prompt reaction
of this substance with so many organic compounds. It
can be computed from graphs like that of Figure 20 by

Fig. 20. Semi-logarithmic plot of survivor curves of Bacterium coH
treated with chlorine. Abscissa: time of exposure; ordinate: per cent
of survivors, on logarithmic scale. (From data of Mallmann and
Audrey, 1940.)

estimating the time required to kill the same percentage
of bacteria by different concentrations. A horizontal line
at 10% survivors gives ditferent results than one at 1%
survivors. The concentrations in this graph are in the
ratios 1 :2 :3 :4 :5 the corresponding times to kill 90% of
the cells are 140, 92, 70, 62, and 22 minutes, and the re-
spective concentration exponents are 1.66, 1.61, 1.69, and
0.87. From the times required to kill 99% of the cells,
the concentration exponents are 2.1, 1.67, 1.70, and 0.93.
The relation between concentration and death time, from

DISINFECTANTS

95

the data of Charlton and Levine on the death of spores
of B. Metieus is shown in Figure 21.

Fig. 21. Relation between chlorine concentration and death time of
spores of B. metiens at various acidities. Abscissa: Concentration of
chlorine, on logarithmic scale; ordinate: time required to kill 99%
of the spores. (From data of Charlton and Levine, 1937.)

Iodine is used extensively in medicine as a disinfec-
tant. As it is almost insoluble in water, it is applied
either in alcoholic solution or in aqueous potassium iodide
solution. Strangely enough, no experiments could be
found which would permit a calculation of the concen-
tration exponent. Nor has any study of the etfect of
acidity on iodine disinfection been made, as far as the
author is aware. Whether iodine acts as such, or as
HOI (corresponding to the HOCl of chlorine) or as KL
in the solutions with potassium iodide, is not known.

8. Soaps. The mechanism of disinfection by soaps
is rather complicated. Not only does their efficiency de-
pend upon two variable factors, namely, the proportion
of the various fatty acids, and the amount of excess
alkali, but they have also a highly selective action, lethal

96 DISINFECTANTS

concentrations occasionally differing a hundredfold be-
tween strains of the same species (Tilley and Schaffer,
1930).

Walker (1924) began to unravel this problem by
studying first the efficiency of the potassium and sodium
salts of each fatty acid; later (1925) he investigated the
various soaps from different fats and oils. Outstanding
among the components studied was Na or K laurate
which killed pneumococci and streptococci in high dilu-
tions, and also affected typhoid bacteria, but had no
action on a strain of Staphylococcus aureus which was
completely resistant to all soaps. Klarmann (1933) ver-
ified the exceptionall}'^ strong effect of lauric acid.

Later, in 1941, Klarmann and Shternow found capric
acid more efficient than lauric. Myristic acid was weaker,
though more jDowerful than palmitic. This held true for
all bacteria tested except for typhoid and dysentery bac-
teria for which palmitic and stearic acids were as toxic
as lauric and myristic.

The salts of unsaturated acids (oleic, linolic and lino-
lenic) are highly bactericidal for pneumococci and strep-
tococci, but entirely harmless for typhoid bacteria. So-
dium ricinoleate has a marked selective action, which is
utilized in the isolation of the colon-typhoid group.

Walker (1925) investigated commercial soaps as well
as some he made himself from pure fats. He too found
the coconut oil soap, which contains much lauric and my-
ristic acid, superior to all others in killing streptococci
and typhoid bacteria (Table 18). In hand-washing, the
soap concentration varies from 0.3% in hasty rinsing to
15% in prolonged scrubbing of the hands, and can be
considered to average about 8% when the hands are wash-
ed thoroughly with the purpose of cleaning them. Thus,
while any ordinary soap is sufficient to destroy any ad-
hering streptococci, pneumococci. and diphtheria bacte-
ria, for typhoid bacteria, coconut oil soap acting for

DISINFECTANTS

97

3 minutes is necessary. At higher temperatures the
efficiency of soaps is considerably increased. Klar-
mann and Shternow (1941) however, found that typhoid
and dysentery bacteria, as well as streptococci, are
not completely killed on infected hands by the cus-
tomary use of soaps in lavatories. But their standard
of time differs from that of the previously mentioned in-
vestigators.

TABLE 18

Concentrations (in per cent) of neutral soaps required to kill
bacteria. (From Walker, 1925.)

Streptococcus

Bacterium

Staphylococcus

hemolyticus

typhosum

aureus

Fat used for

20°

20

0

350

20°

350

soap

21/2

15

21/2

15

15

15

15

min.

min.

mm.

min.

mm.

min.

min.

Coconut oil

(Na)

0.31

0.16

2.5

1.25

0.62

>10

>10

Beef tallow

(K)

0.62

0.16

>10

10.0

0.62

>10

>10

Olive oil (Na)

0.62

0.16

>10

>10

1.25

>10

>10

Cottonseed oil

(Na)

0.31

0.16

....

5.0

>10

Linseed oil

(K)

0.62

0.31

>10

10.0

0.62

>10

>10

Resin (Na)

0.62

0.62

10.0

2.5

0.62

2.5

1.25

Walker's results were verified by Tilley and Schaffer
(1925) and by Schaffer and Tilley (1930) who, further-
more, showed that a slight increase in alkalinity in-
creased the germicidal action. The effect of organic
matter was studied by adding 50% skim milk which de-
creased the efficiency of all soaps greatly. According to
Eggerth (1927), serum and other proteins decrease the
efficiency of soaps while certain lipoids increase it.

Schaffer and Tilley (1930) tested also the effect of ad-
ditions of phenol, cresol and other disinfectants to soaps

98

DISINFECTANTS

while Eggerth studied the soaps from substituted fatty
acids. He found that oc-brom fatty acids are more ger-
micidal than the unsubstituted acids (1929), that ct-hy-
droxy acids are highly germicidal to some species, but
that the substituted compounds have become even more
selective than the regular soaps (1930) ; a similar in-
crease in selectivity was observed with a-mercapto and
a-disulfo soaps (1931).

Of greatest interest to the mode of action of fatty
acids and soaps is the etfect of pH which had been em-
phasized by Eggerth in 1927. Klarmann (1933) meas-
ured the amount of various acids required to kill Staphy-
lococcus aureus at 37° in 2 hours at different pH values.
His results are shown in Fig. 22.

Fig. 22. ■ Concentration of various fatty acids required to kill Staphy-
lococci in 2 hours, at different acidities. Abscissa: pH; ordinate: per
cent concentration of the fatty acid, on logarithmic scale. Ce: caproic
acid, Ce: caprylic acid, Cio: capric acid, C12: lauric acid, C14: myris-
tic acid, Cis palmitic acid, Cis: stearic acid. (From data of Klarmann,
1933.)

DISINFECTANTS 99

The lower acids, up to 10 carbons, are efficient only
in acid media, at pH 6 or less. In this range, the dis-
infecting power increases with the molecular weight up to
C12H24O2, lauric acid. Myristic acid, C14H28O2, is not
quite as active in acid media, and the higher acids have no
disinfectant properties at all. The fact that the effi-
ciency increases with the acidity indicates that death is
brought about by the undissociated part of the molecule.
(For a discussion of this point and for the computation
of the undissociated fraction see p. 157). The dissocia-
tion constant is 1.8 X 10 '^ for acetic, 1.5 X 10"^ for butyric,
and 1.4 X 10'^ for caproic and caprylic acid. Klar-
mann's data show that death is brought about by 40-60
mg. undissociated caproic acid in 100 cc. Assuming 1.4
X 10^ as dissociation constant for all the higher homo-
logues, the corresponding values for the higher acids are
20-40 mg. for Cs, 1-1.5 mg. for Cio, 0.2 - 0.8 mg. for C12
and about 2 mg. for Cu. Palmitic and stearic acid are
so insoluble (0.00092 and 0.00038% respectively at 37"^
according to Ralston and Hoerr, 1942) that they could not
be expected to act as disinfectants.

At neutrality, capric, lauric and myristic acids are the
only acids capable of killing staphylococci. The lower
acids are practicallj^ completely dissociated, and the high-
er acids remain insoluble. But with increasing alkalin-
ity, these latter dissolve, and at pH 9 to 10, palmitic and
stearic acid become weak disinfectants and myristic acid
becomes stronger again, resembling the higher acids,
w^hile capric acid loses its power entirely.

The detergent properties of soaps have been sur-
passed by those of new synthetic detergents developed
during the last decade. These so-called ' * wetting agents ' '
are discussed here in connection with the soaps not only
because they are substitutes for soap, but also because
many of them contain fatty acids. The ''cationic de-
tergents" have the fatty acids bound on ammonium de-
rivatives in the cationic part of the molecule, the "ani-

100 DISINFECTANTS

onic detergents" contain them, often in form of sulfonic
acids, in the anion.

Many of the wetting agents are not only detergents,
but good disinfectants. Baker, Harrison and Miller
(1941 and 1942) found that cationic detergents as a
group kill Gram-positive bacteria more readily than
Gram-negative species. Of the 9 detergents investi-
gated, all but one killed streptococci, lactobacilli and
staphylococci within 10 minutes when diluted 1:3000.
Two of the 9 failed to kill Bacterium coli and typhosum,
three failed with Proteus vulgaris.

The anionic detergents were much less efficient. Eight
of them were tested with the bacteria mentioned above, in
1:1000 dilution. After a 10 minutes' exposure, only
three had killed the lactobacilli, two had killed the strep-
tococci, and none was powerful enough to kill the two
staphylococci; even with 90 minutes exposure, only one
of the eight detergents could destroy the staphylococci.
None of these compounds was effective against any of the
Gram-negative test organisms.

A very remarkable phenomenon was observed by Mil-
ler, Abrams, Huber and Klein (1943) when cationic de-
tergents were used to disinfect hands. It seems that the
detergent deposits on the hands an invisible, non-percep-
tible film which retains the bacteria under it without
killing them, while the outer surface becomes sterile.
The film is not easily broken mechanically, c.g,, by scrub-
bing, and hands treated with cationic detergents can be
considered practically sterile for a considerable time.
Frequent dipping into saline solution Or long rinsing
with running water tends to deteriorate the film. Soap
breaks it up promptly, releasing the viable bacteria un-
derneath. The authors did not offer an explanation for
the different action of the two sides of the film.

DISINFECTANTS IQI

VII. STANDARDIZING OF DISINFECTANTS

The thorough quantitative studies of disinfectants be-
tween 1890 and 1900 made it possible to compare their
eniciency, and Rideal and Walker in 1903 proposed to use
phenol as the standard of comparison because it is a
reliable, not very selective disinfectant that can be ob-
tained chemically pure at reasonable cost anywhere.
Their method consisted in inoculating various dilutions
of phenol, as well as of the disinfectant under test, with
equal amounts of bacteria and determining the death
times after transfer into broth. Dilutions of equal death
times were compared and their ratio was believed to rep-
resent a reliable index of the relative disinfectant actions.
Thus a '^ phenol coefficient" was obtained which told how
many times more efficient the substance investigated was
than phenol.

This method received a severe jolt when Chick and
Martin (1908) showed that the phenol coefficient of sev-
eral common disinfectants varied very much when the
death time was changed (see Table 19). Watson (1908)
explained this variation quite simply. Chemical reac-
tion rates are not always proportional to the concen-
tration, but to the concentration raised to a certain
power, and, with phenol, the concentration exponent is
about 6. No other disinfectant except the homologues
of phenoP has such a high exponent (a list of concentra-
tion exponents is given in Table 15). The action of most
disinfectants is approximately proportional to their con-
centration, and the decision to use phenol, with its ab-
normal concentration exponent, as a standard, was very
unfortunate.

Phelps (1911) pointed out that to characterize a dis-
infectant, we need only to know the death rate and the

1. Concerning the exceptional values for alcohols see p.

68.

102

DISINFECTANTS

concentration exponent. They are the only unknowns
in the equation

Ktc" = log

initial number
survivors

The values of K and n characterize the disinfectant far
better than the phenol coefficient, and if the temperature
coefficient were also known, the effect of the disinfectant
would be known for any combination of concentration,
exposure-time and temperature.

TABLE 19

Variation of the phenol coefficient of mercuric chloride with
different death times. (From Chick and Martin, 1908.)

Disinfectant

Death time
in minutes

Concentration (%)

required to kill In

that time

Phenol
coefficient

Phenol

Mercuric chloride

Phenol

Mercuric chloride

Phenol

Mercuric chloride

2.5
2.5

10
10
30
30

1.20

0.008

1.04

0.006

0.990

0.0018

13.6
173
550

Phelps himself did not continue these studies and
did not attempt to show with experimental data the
practicability of his proposal. In fact, this proposal was
never given a thorough test. The idea was dismissed on the
ground that the deathrate constant K fluctuates enor-
mously. The data obtained by Chick are usually given
as proof of that inconsistency. The fluctuations are
quite large indeed, even when the test is made with the
same strain of bacteria, grown in the same medium, at
the same temperature, and to the same age. Despite
all attempts at uniformity, Chick obtained on different
days the following deathrate constants for Bad. paraty-
pJiosum exposed to 0.6% phenol at 20°C: 0.19, 0.45, 0.25,
0.82, 0.27.

DISINFECTANTS 103

If such fluctuations occurred in the same laboratory, the
results from different laboratories might be expected to
vary still more. Accordingly, the deathrate constant was
generally considered unreliable, and the concentration
exponent never became popular. Standardization was
continued by the phenol coefficient. The method was
''improved" by detailed specifications as to medium,
time, temperature, bacterial species and strain to be
used, and as to the size of the loop employed for its
transfers.

These specifications limit the usefulness of the phenol
coefficient because the values thus obtained are correct
only in this narrowly limited set of specified conditions
and may be of little use with changed conditions. But the
limitations made the phenol coefficients more uniform.

The present Standard Method in the United States is
that outlined by the Federal Drug Administration, usually
referred to as the F.D.A. method. In order to compute
the phenol coefficient of an unknown disinfectant, the low-
est concentration of the latter is determined which will
kill the test culture in 10, but not in 5 minutes. At the
same time, the lowest phenol concentration is ascertained
which kills the bacteria in 10, but not in 5 minutes. The
ratio of these two concentrations is the phenol coeffi-
cient, which tells how much stronger the unknown dis-
infectant is than phenol.

In the above example. Table 19, the concentrations which
kill in 10 minutes are 1.04% for phenol and 0.006 for
HgCL. The phenol coefficient is therefore 1.04/0.006 =
173. The other two values of Table 19, which show the
effect of concentration, are disregarded in the definition
of the phenol coefficient which is considered correct only
for 10-minute exposures, and not for 2.5 or 30 minutes.

While the very important role of concentration is com-
pletely neglected by the F.D.A. method, attention is given
to the variability of the test culture. The ''Hopkins

104 DISINFECTANTS

strain" of the typhoid bacterium is the only strain used
for standard experiments, but even with this strain death
may occasionally occur more rapidly or more slowly than
usual. In order to exclude such fluctuations, the F.D.A.
method considers phenol coefficients as being correctly
established only when the control cultures are killed by
phenol within the following ranges of time and concen-
tration :

Dilution Concentration 5 min. 10 min. 15 min.

1:90

1.11%

+

0

0

1:100

1.00%

+

-i-

+

and

1:90

1.11%

0

0

0

1:100

1.009c

+

+

0

This means that the culture may be so resistant that it
can tolerate 1.00% phenol for 15 minutes, but must
be killed by l.ll%o in 10 minutes, or it may be so sensi-
tive that it is killed by 1.11% in 5 minutes, but it must
survive for at least 10 minutes at 1.007^.

It may seem strange that the death rates fluctuate so
greatly while phenol coefficients are uniform enough to
be universally accepted. However, the fact is that we
are really fooling ourselves about the accuracy of the
phenol coefficient. Both methods are based on the same
principle, and their accuracy should therefore be ap-
proximately the same. The apparently greater accur-
acy of the phenol coefficients is due to the enforcen-ent
of very specified conditions, and to the discarding of all
those results where the controls did not remain within
the narrow limits permitted by the rules of the method.
If such severe restrictions had been applied to the de-
termination of the concentration exponent, its accuracy
would have been equal to that of the phenol coefficient.

But even the range allowed by the F.D.A. method
permits a great fluctuation of results; an experiment is

DISINFECTANTS 105

considered equally correct whether the bacteria are
killed by 1.11% phenol in 14 minutes or by 1.0% in 11
minutes. This means a fluctuation in the deathrate
constant the range of which can be calculated. In our
standard equation

Ktc" = log ii^itial number
survivors

the ratio of the initial number of bacteria to that of the
survivors is constant for any set of experiments and
therefore, for any combination of time and concentra-
tion. We have, then,

Substituting the above limiting values of the F.D.A.
method for t and c, and taking 6 as the concentration
exponent of phenol, we find

Ki X 11 X 1.0« = K2 X 14 X l.ir
and, Ki=2.4K2.

The deathrate constant in one experiment may be 2.4
times as high as in the other.

We must further consider that the killing time by the
unknown disinfectant must be more than 5 and less than
10 minutes. Sterility may be reached in one case in 6
minutes, in another case in 9 minutes. This increases
the range of the deathrate by 9/6 or 1.5, so that in two
standard experiments, one may have been carried out
with a deathrate constant which was 1.5 X 2.4 = 3.6
times as large as in the other experiment. Any experi-
ment having a greater error, i.e., greater than 360%, is
discarded as not being carried out under standard con-
ditions. This is the error which the specified conditions
permit and to which must be added the personal error
of the experimenter.

That phenol coefficients are by no means precisely de-
fined values may be seen from Table 20, which shows that
two different authors obtained for mercurochrome phenol

106

DISINFECTANTS

TABLE 20

Phenol coefficients

(The data of this table are gathered from McCullough's book 1936,
except the first five figures in column I for Staph, aureus which are
from Thomas, 1932, and those of column II which are from Birkhaug,
1933.)

Disinfectant

Bacterium

Staphylococcus

typhosum

aureus

I

II

Hydrogen ions

1,400

Hydroxyl ions

23,000

Bichloride of mercury

775

166

189

Merthiolate

600

62.5

1412

Metaphen

1,792

2250

1647

Mercurochrome

133

267

90

Ethyl mercury chloride

88.3

233

Phenyl mercuric nitrate

2259

Silver nitrate

173

Potassium permanganate

200*

Hydrogen peroxide

5.6

"4.8

Chlorine

218-233

220

Iodine

170-235

170-235

Methyl alcohol

0.026

0.030

Ethyl alcohol

0.040

0.039

n-Propyl alcohol

0.102

0.082

n-Butyl alcohol

0.273

0.22

n-Amyl alcohol

0.78

0.63

n-Hexyl alcohol

2.3

n-Heptyl alcohol

6.8

n-Octyl alcohol

21.0

Phenol

1.0

1

o-Cresol

2.2

2.3

p-Cresol

2.4

p-Xylenol

5.5

Thymol

28.5

23""

Carvacrol

27.5

o-Phenyl phenol

50

57

p-Benzyl phenol

62

Hexyl resorcinol

140

72

Formaldehyde

1.05

Acetaldehyde

0.09

■

Propionic aldehyde

0.15

Butyric aldehyde

0.32

Methyl methyl ketone

0.04

0.036

Methyl ethyl ketone

0.102

0.080

Methyl n-propyl ketone

0.275

Methyl n-butyl ketone

0.78

....

Methyl n-amyl ketone

2.3

....

*)With washed cells

DISINFECTANTS

TABLE 20 (Continued)
Phenol coefficients

107

Disinfectant

Bacterium
typhosum

Staphylococcus
aureus

n-Propyl amine

1.6

0.23

n-Butyl amine

2.1

0.52

n-Amyl amine

3.0

1.3

n-Hexylamine

4.5

3.1

n-Heptylamine

6.5

....

Diethylamine

2.2

0.21

Di n-propylamine

2.0

Di n-butylamine

4.7

Triethylamine

1.5

o!25

Benzylamine

1.0

Aniline

0.57

0.50

p-Toluidine

1.25

Methyl aniline

1.8

Ethyl aniline

3.0

....

p-Chlorophenol

4.3

o-Methyl chlorophenol

12.5

o-Ethyl chlorophenol

28.6

0- n-Propyl chlorophenol

93.3

0- n-Butyl chlorophenol

141

0- n-Amyl chlorophenol

156

0- sec. Amyl chlorophenol

46.7

0- n-Hexyl chlorophenol

(23.2)

o-Cyclo-hexyl chlorophenol

<26.7

0- n-Heptyl chlorophenol

(20.0)

coefficients of 90 and 267 respectively, one value being
3 times as large as the other. The phenol coefficient of
chlorine was found by McCulloch (1936) to be 139, by
Rideal 218 and by Zoller 223, all data being obtained with
Bacterium typhosum.

The above-mentioned deathrate constants observed by
Chick for the effect of phenol upon paratyphoid bac-
teria show a deviation of the extremes greater than 360%,
Of the 5 constants found on different days, 4 differ by
less than 360% but the highest value is 4.3 times as large
as the lowest while the F.D.A. method considers 3.6 times
the permissible limit. Obviously, the culture was unus-

108 DISINFECTANTS

ually sensitive on this particular clay. The F.D.A.
method discards such data as abnormal, and they do not
appear in the reports. But Chick's unexpurgated data
are used as proof that deathrate constants fluctuate too
much to be of value. The establishment of limits of sen-
sitivity for the phenol coefficient proves clearly that the
same fluctuations occur there, but they are never men-
tioned in publications. As has been said before, if we
consider the phenol coefficients more accurate than the
deathrate constants, we are merely fooling ourselves.
They appear more accurate only because the larger de-
viations are not published.

As a result of this anathema on the deathrate constant,
few data can be found in the literature. Watkins and
Winslow (1932) give four sets of data on the death of
Bacterium coli by alkali. Each set concerns a different
cell age and consists of 4 single experiments. The high-
est and lowest values of K in each set are:

0.119 and 0.166, the highest being 40% larger than the lowest

0.136 and 0.191, " " " 40%

0.199 and 0.242, " " " 22%

0.336 and 0.478, " " " 42%

This proves that with our present knowledge of the in-
fluence of pH, organic matter, cell concentration, culture
medium and age of the cells, which was not available in
1911 when Phelps proposed the standardization of disin-
fectants by death rates, the deathrate constants can be
measured as accurately as the phenol coefficients. They
would have the advantage of telling the efficiency of a dis-
infectant directly, and not merely from comparison with
another disinfectant such as phenol.

Finally the attention of experimenters should be called
to a common source of error in any method which in-
volves the determination of a death time. It is not al-
ways realized that a culture showing no growth when
transferred to broth may not be sterile. ''No Growth"
means that no living cell remained in the one loopful

DISINFECTANTS

109

transferred, but the latter amounts to only 0.01 cc, so
that ''no growth" indicates that the culture contained
less than 100 surviving cells per cc. It may have contained
50. If the average inoculum for the determination of the
phenol coefficient, 100 million cells, is reduced to 50 living
cells per cc. in 10 minutes, it requires nearly 5 minutes
longer to make the entire culture sterile, i.e., to reduce
the survivors to 0.1 per test tube of 5 cc.

Other Methods of Standardisation.
Another method suggested by Bronfenbrenner, Hershey
and Doubly (1938) for the standardization of disinfect-
ants should be mentioned here. These authors, working
on the theory that death is due to inactivation of enzymes,
measured the etfect of disinfectants upon the oxygen up-
take of Bacterium coli, and of liver cells (see Table 21).
They compared the concentrations causing an equal de-
crease in oxygen consumption, and thus obtained a tox-
icity index comparable to the therapeutic index. In 1939,
Ely tested this method on several disinfectants, but his
results were not very encouraging.

TABLE 21

Manometric evaluation of disinfectants by the determination of
the decrease in oxygen uptake. (From Bronfenbrenner, Hershey and
Doubly, 1938.)

Disinfectant

Concentration (%) which

reduces oxygen uptake

to one-half

•

Toxicity
index

Liver cells

Bad. coli

Phenol

Tincture of iodine

Mercuric chloride

0.23

0.0165

0.0040

0.26

0.0425

0.00085

1.1
2.6
0.21

PROPOSAL OF A NEW METHOD FOR TESTING DISINFECTANTS

^ It is possible by relatively slight changes in the tech
nique to measure the actual efficiency of a disinfectant a

/

'^^

f^^^

110 DLSIXFECTAXTkS

different concentrations instead of its relative value in
comparison with phenol, and also to measure the effect of
organic matter. Such a method requires more time, but
the far more useful result makes this expenditure of time
worth while. The method consists, in principle, in deter-
mining the time in which the standard bacterial suspen-
sion will be sterilized by various dilutions of the disin-
fectant.

In the method of the phenol coefficient, the effect of
organic matter is not sufficiently eliminated, since 10% of
the suspension medium is broth. In the present method,
the bacteria are grown in broth in a centrifuge tube
which is centrifuged when the culture is 24 hours old,
and the cells are then re-suspended in sterile w^ater or in
a mineral solution known to prolong their life without
providing food. (Such solutions have been worked out
by Zeug, 1920.) The cell concentration is the same as it
was in the culture.

Cultivation of the test organism and all other details of
the procedure are the same as for the determination of
the phenol coefficient. The disinfectant is made up in va-
rious concentrations (the pharmacologists will probably
continue to prefer "dilutions" to concentrations) and the
death time is determined as usual by transfer with a
standard loop into broth tubes. The culture may also be
tested against phenol so as to check unusual fluctuations
of the test organism.

The results can be arranged in the form of a chart as
in Table 22 or in the form of a graph in which the death
times are plotted against concentrations on a double loga-
rithmic scale, as in Figure 23. The tabular form is handy
for the recording of data for further reference, but the
graph is far more instructive. Since the death time is
not known precisely, but lies within a certain range, the
entire range should be plotted on the graph as a line, and
not mer9ly its average (see Figures 17 and 23). Thus in-
stead of drawing the disinfection curve through points,

Dh'^INFECTAXTl^ 111

we draw them through lines which represent ranges of
death times.

The curve obtained must be a straight line as explained
on p. 72. Different slopes indicate different concentra-
tion exponents. Figure 23 shows a number of such curves
for common disinfectants. Phenol and permanganate
furnish slopes quite different from those of the other
disinfectants, thus indicating unusual concentration expo-
nents.

Rectilinear interpolation and extrapolation give the
death times for any concentration, or the concentrations
necessary to kill the bacteria at any exposure time.

These curves represent the effect of the disinfectant
upon bacteria in the absence of organic matter. The ef-
fect of organic matter can be measured quantitatively by a
parallel set of death time determinations, the disinfectant
being dissolved in water containing 10% skimmed milk.
The results are plotted in the same way as those with
water, and again, straight lines are obtained.

The effect of organic matter is expressed by the ''per-
centage delay." This is calculated from the two death
times at the same concentration (which may eventually
be obtained by extrapolation). Death time in milk will
be longer than death time in water or salt solution, and
the percentage of increase in time tells to what extent or-
ganic matter interferes ; e.g., for hydrogen peroxide the
delay in skimmed milk is 30%, for formaldehyde 20-27%,
for H^CL many thousand per cent.

The phenol coefficient can be obtained directly from the
graph by dividing the phenol concentration which kills
"between 5 and 10 minutes" by the corresponding con-
centration of the disinfectant under test. In most cases,
the coefficient measured at 5 minutes will be larger than
that measured at 10 minutes because, phenol having the
highest concentration exponent known, the slope of the
phenol disinfection curve is more nearly vertical than

112

DISINFECTANTS

TABLE 22
Death times (in minutes) of Bacterium coli suspended in water and in milk

Concentr.

(%):

Dilution: 1:

2 1.5 1.0 0.8 .06 0.4 0.2 0.1
50 67 100 125 167 250 500 1000

0.08 0.06 0.04 0.02 0.01
1250 1670 2500 5000 10,000

Phenol

In water
In milk

1 13 90

25 95 395 1440

Aniline

In water
In milk

6 40

Formaldehyde

In water
In milk

5 9 12 18
7 12 17 30

Hydrogen Peroxide

In water
In milk

2 3 5 8 24 65
3.5 5 7.5 13 35 90

Mercurochrome

In water
In milk

8 20 50

5 15

65 90

Potassium Permanganate

In water
In milk

0.8

In water
In milk

In water
In milk

DISINFECTANTS

113

TABLE 22 (Coutinued)

8 ppm. 6 ppm. 4 ppm. 2 ppm. 1 ppm.
125.000 167,000 250,000 500,000 1,000,000

0.008 0.006 0.004 0.002 0.001
12,500 16,700 25,000 50,000 100,000

Mercurochrome

20

30

45

Potassium permanganate

2.4

45

Mercuric Chloride

15

1.2
38

1.8
52

18

65

Silver Nitrate

25

90

MINUTES

8 ip 12 16 20 25 30 40 60

M 000 000

0^ 1 ~2 3 4 5 6 8 10 12 16 20 25 40 60 80 120

Fig. 23. Relation between death time of Bacterium coU and concen-
tration of disinfectants. The heavy horizontal bars (representing ranges
of death time) and the solid lines which join them refer to washed
bacteria, the multiple horizontal bars and the broken lines which
join them refer to suspensions containing 10% skim milk. Abscissa:
death time, on logarithmic scale; ordinate: concentration of disin-
fectant, on logarithmic scale.

DISINFECTANTS 115

that of other disinfectants. (As was said above, this ac-
counts for the great discrepancies in the values of the
phenol coefficient determined by different workers for a
given disinfectant.)

The disinfection curves for different temperatures
will be parallel lines equidistant for equal temperature
changes as long as the temperature coefficient is constant.

The efficiency of a few disinfectants, such as chlorine
and mercurochrome, varies greatly with the acidity of
the test solution. This effect can be measured by draw-
ing separate disinfection curves for each pH that is of
interest. The lines will be jDarallel, as a rule, and their
distance measures the effect of pH on the disinfecting
power.

Justification of the N e iv Features
of the Method. The advisability of measuring
the death times for several different concentrations is
quite evident from the different slopes of the disinfec-
tion curves of the various disinfectants. These different
slopes indicate how differently the efficiency of the disin-
fectants is affected by changes in concentration.

For the determination of the effect of organic matter,
skimmed milk seems ideal because its composition is re-
liably uniform all over the world, and because it contains
dissolved proteins and carbohydrates as well as finely
suspended organic matter (casein).

The resuspension of centrifuged bacteria in water is
necessary to remove the broth which contains sufficient
organic matter to interfere with such strong disinfect-
ants as HgCL or iodine. In the F.D.A. method for the
phenol coefficient, the Vii cc. of culture used brings 1500
ppm. of organic matter into the disinfectant. One cen-
trifugation reduces it to about 25 ppm. dissolved organic
matter besides 20 ppm. of bacterial solids. A second
washing (which is advisable with such disinfectants as
chlorine and iodine) reduces it to about 0.5 i^pm.

116 DISINFECTANTS

The use of pathogenic species for disinfection studies
'offers no advantages. Saprophytes can be handled more
easil}^ The strain of Bacterium coli used by us has a
phenol coefficient almost identical with that of the Hop-
kins strain of Bad. typhosum.

DETAILS OF THE PROCEDUEE

It is advisable to make a strong solution of the disinfectant
some time before the test, to be certain of a sterile start. Di-
lutions are made, as was said above, either with water, or with
a mineral solution which keeps bacteria viable for a long time.
For Bacterium coli we used a solution containing 0.05% NaCl,
0.02% KCl, 0.05% CaClo and 0.05% MgCl,, although Brooks
and Winslow (1927) have shown that this species can remain
alive in distilled water for a long time. All dilutions of dis-
infectants, and all salt solutions must be prepared with glass-
distilled water. Small Erlenmeyers containing about 50 cc.
of this salt solution, and others containing skimmed milk di-
luted 1 :10 are sterilized and kept in stock.

The test culture to be used is a 21-hour culture grown in
standard nutrient broth at 37°. It is ceutrifuged, and the bac-
teria are re-suspended in the sterile salt solution to the volume
of the original culture.

The actual exposure of the bacteria to the disinfectant is
done in 18 mm. test tubes which permit thorough mixing. The
final volume is always 5 cc. If the concentrations 0.2%, 0.1%
and 0.05% are desired, they can be made from a 0.5% solution
of the disinfectant according to the following chart:
0.2 % = 2 cc. of 0.5% + 2.5 cc. salt sol'n -f 0.5 cc. bact. susp'n
0.1 % = 1 cc. of 0.5% -j- 3.5 cc. salt sol'n + 0-5 cc. bact. susp'n
0.05% = 0.5 cc. of 0.5% + 4.0 cc. salt sol'n + 0.5 cc. bact. susp'n

The disinfectant and salt solution (or milk) are mixed first
and then the tube is shaken and placed into a water bath at
20°. When it has reached a constant temperature, the bac-
terial suspension is added. The probable death times for each
concentration are estimated, and arranged into a time sched-
ule of the following type :

0.2%: 1/2, 1, 2, 3, 5 minutes;
0.1%: 2, 4, 8, 15, 30 minutes;
0.05%: 15, 30, 45, 60, 90 minutes;

DISINFECTANTS 117

or preferably plotted in a graph which tells what is to be done
every minute, e. g.,

0 1 2 3 4 5 G 7 8 9 10 11 12 13 14 15

X — o — —

x-o-o — o — o — o

X o o o

where x indicates inoculation and o transfer of a loopful to
new broth.

Thus only the disinfectant concentration of 0.05% is inocu-
lated at the start, the inoculation of the 0.2% concentration
comes one minute later, and that of the 0.1% solution not un-
til 5 minutes later. It is usually possible to work out a sched-
ule by which a large number of dififerent determinations can
be fitted together so as to avoid long waiting between transfers.

With an entirely unknown substance, a preliminary test
will save time. A spacing of the concentrations in decimal
multiples such as 1%, 0.1% and 0.01%, or 400 ppm., 40 ppm.,
4 ppm., and a spacing of the times in multiples of their log-
arithms are advisable. A good range of preliminary testing
times is that which includes 1, 3, 10, and 30 minutes, since
their logarithms are about 0.5 apart (0; 0.48; 1.0; 1.48). If
the results of such a preliminary test are entered into a double
logarithmic graph, the slope of the resulting curve tells at once
the more desirable combinations of times and concentrations to
be used. If, for example, one concentration is strong enough
to kill all bacteria in 1 minute while one-tenth of that concen-
tration does not kill in 30 minutes, this indicates a high con-
centration exponent, which is a warning not to space the con-
centrations too widely in the final test; in this case, a range
of 1%, 1.5% and 2% may suffice. Phenol, which at 2% kills
in less than i/o minute and at 0.2% requires more than 10 hours,
and KMnO^, which at 100 ppm. kills in less than 1 minute and
at 10 ppm. requires 10-20 hours, are typical examples of disin-
fectants with very high exponents. A glance at Figure 23 will
help in deciding the most appropriate range of dilutions to
be used.

In order to prove that the test culture has a normal sensi-
tivity, the author has always determined simultaneously its
death time in a 1% phenol solution.

II. ANTISEPTICS

Under this general term are assembled those com-
pounds which are used primarily to prevent multiplica-
tion of bacteria but are not intended to produce com-
plete sterility. While disinfectants, as was already said
above, disrupt the reproduction mechanism irreversibly
and irreparably, antiseptics prevent multiplication only
as long as they are present; their removal permits the
various cell mechanisms to resume their normal function.
It was also pointed out that antiseptics often kill some
or even many of the cells, but death then occurs at such
a low rate that it would take weeks or even months to
achieve complete sterility. In some cases, this slow death
occurs only at the start and is followed later by a period
of no further dying, and sometimes even by multiplica-
tion.

The practical efficiency of antiseptics can be measured
only in one way, namely by determining the lowest con-
centration which prevents multiplication. The inhibit-
ing dose may vary not only with the medium and the
temperature, as would be expected from analogy with
disinfectants, but also with the number of cells present.
For example, the multiplication of Bacterium coll in
broth is completely inhibited by 1 :100,000 crystal violet
if only 100 cells per cc. are present; but with 10,000
cells per cc, multiplication sets in after a temporary
delay. That delay in multiplication does not occur with
disinfectants; with them, when larger inocula require a
longer exposure time\ it is to cause death, not to retard
multiplication.

1. See p. 54.

ASThSEPTICS 119

An outstanding feature of the antiseptic reaction is its
immediate effect. As soon as the cells are in contact with
the antiseptic, they cease to multiply. The reaction may
be compared to pulling the switch of an electric motor.
The motor stops and continues to be motionless as long
as the current is interrupted, although the motor is
intact. It runs again when the switch is thrown in, just
as the cells begin to multiply again when the antiseptic
is removed.

This immeasurably rapid reaction of the antiseptic is
different from the slow death-producing reaction char-
acteristic of a disinfectant. To continue our simile of the
electric motor, the disinfectant action is comparable to
a gradual deterioration of the motor, as through rust or
corrosion by acid vapors. This deterioration also inter-
feres with the running of the motor, and it is irreversible,
but deterioration is slower than pulling a switch.

The slow reaction of the disinfectant requires a con-
siderably higher concentration of the toxic substance
than the instantaneous but reversible reaction produced
by the antiseptic, and, in fact, all disinfectants act also
as antiseptics at low concentrations. They inhibit mul-
tiplication at once, but produce their irreversible effect
only gradually. They first, so to speak, pull the switch
of the motor, and then cause it to deteriorate. The an-
tiseptic only pulls the switch. If there is also deteriora-
tion, it is very slow.

The essential differences between the antiseptic and
the disinfectant reactions are thus that one is slow and
gradually progressing, but requires a fairly high concen-
tration of the toxic agent, and produces an irreversible
effect while the other is practically instantaneous and re-
quires a relatively dilute solution, but its effect is re-
versible. These reactions probably affect different cell
constituents, or at least different parts of the molecule
(^f the same constituent. A good example of the two re-

120 ANTISEPTICS

actions produced by the same toxic agent lias already
been given in the experiments by Gegenbauer with mer-
curic chloride. The inhibition of growth of staphylococci
could be reversed during the first hour of exposure by re-
peated washing, but after longer exposure, even an anti-
dote would not restore the ability to multiply. Another
example in which antisepsis and disinfection effects have
been measured separately will be found later in the dis-
cussion of crystal violet.

Antiseptics cannot be studied in the same way as dis-
infectants because with them there is no measurable rate
of reaction, no death rate and no death time. The in-
stantaneous reaction is followed by a period of no fur-
ther change in the status of antisepsis. Even if some or
all cells slowly die, that does not affect the status of
antisepsis, for the dead cells continue not to multiply, just
as the living cells do not multiply. The exceptional
case of the sulfa drugs which permit a short period of
apparently normal growth before the status of antisepsis
becomes established will be treated in a separate chapter.

An antiseptic is characterized by the lowest concen-
tration which will inhibit multiplication. Higher con-
centrations produce no effect different from that of the
sufficient concentration. Since little can be concluded
from the knowledge of only the inhibiting concentra-
tion, the effect of more dilute solutions is also studied.
Such solutions are not truly antiseptic, but important
conclusions can be dra\vn from their effect on the rate of
multiplication, the length of the Jag period, and the final
crop. The rate of multiplication is usually retarded, the
lag period is greatly prolonged by some antiseptics, but
not changed at all by others, and the final crop usually
decreases as the antiseptic concentration increases. This
latter effect is well illustrated by the following data on

ANTISEPTICS 121

the influence of various concentrations of sulfanilamide
on a culture of Bacterium coll in broth:
Concentrations Final crop

1.0% < 5,000,000 cells per cc.

0.75% 110,000,000

0.50% 250,000,000

0.25% 900,000,000

0.10% 1,000,000,000

The culture with 1% sulfanilamide showed no visible tur-
bidity, gave a slight indication of growth in the nephelo-
meter, and the plate count definitely proved a slight
multiplication in the course of 10 days, which did not
progress further during another w^eek of observation.

I. TEMPERATURE AND ANTISEPSIS

With disinfectants, an increase of temperature in-
creases the rate of reaction between cells and the toxic
compound, and thus increases the efficiency of the disin-
fectant. Since antiseptics react with immeasurable vel-
ocity, this rate cannot be influenced to a measurable de-
gree by temperature. However, the effect is reversible
and must be considered as representing an equilibrium.
Equilibria are affected by changes in temperature, and
equilibria of instantaneous ionic reactions are as de-
pendent upon temperature as those of slow organic re-
actions.

The equilibrium constants of exothermic and endo-
thermic reactions are affected in opposite ways by a
change of temperature. The precise nature of the reac-
tion which interferes with multiplication when the anti-
septic is added to a culture is not known. It may be
endothermic with one antiseptic and exothermic with an-
other. It is therefore impossible to predict in a general
way the effect of temperature on all antiseptics.

122 AXThSEPTICS

If the antiseptic reaction is endothermic, an increase
in temperature should decrease the efficiency of the toxic
compound. This is contrary to all experience in disin-
fection, but at least one such case has been reported.
Cameron (1930) observed that thermophilic sporeform-
ers causing "flat sours" in canned vegetables required
10 ppm. of gentian violet at 55° C for complete inhibition,
while at 37°, 1 ppm. sufficed to inhibit growth.

As a rule, however, a higher temperature makes an an-
tiseptic more efficient, which suggests that the antisep-
tic reaction is generally exothermic. More antiseptic
will then be needed at low temperatures than at high tem-
peratures to suppress microbial development. McCul-
loch (1936, p. 228) mentions that a certain glue was pro-
tected against molding by a phenol concentration which
was quite sufficient in summer, but which did not pre-
vent spoilage when the temperature dropped to about
10°C.

As no rates can be measured in antisepsis, no temper-
ature coefficients can be computed. The only possibility
of expressing temperature relations is the recording of
the lowest inhibiting concentration for each tempera-
ture.

But here, a considerable range of errors may be ex-
pected. Limiting values cannot be very precise when
the approach to the limit is asymptotic as in antisepsis.
For instance, Cruess and Eichert (1929) report that at
pH 3.0, Saccharomyces ellipsoideus was delayed by 0.02%
sodium benzoate, but began to grow after 12 days ; 0.06%
benzoate inhibited growth completely. The minimal in-
hibiting dose lies somewhere between these two values.
When the limiting concentrations at tw^o different tem-
peratures are determined, and when each value is sub-
ject to a considerable range of error, the ratio of these
two concentrations is no sound basis for an attempt at
explaining temperature effects.

AXTISEPTICS ] 23

II. SELECTIVE ACTION OF ANTISEPTICS

At the beginning- of the discussion of chemical dis-
infection, it was pointed out that the strong disinfectants
kill different species of bacteria at nearly the same rate,
while dyes, benzoic and salicylic acid, and a number of
therapeutic agents show marked differences in their ef-
fects upon different groups of bacteria. Most of the se-
lective compounds must be considered as antiseptics
rather than as disinfectants. Selectivity indicates that
the cell mechanism upon which these substances act dif-
fers in those species which show different resistance.

The most commonly claimed instance of selectivity is
that between Gram-positive and Gram-negative bacteria.
Such correlation between staining property and vital
function would be a very great help to the biochemist.
However, a closer analysis shows that many such claims
are generalizations from tests with only very few species,
and are by no means based upon a representative selec-
tion of Gram-positive and Gram-negative genera. If
the tests are made only with Staphylococcus aureus and
Bacterium typliosum, the two officially designated test
species for disinfectants, no conclusion can be derived
for other Gram-positive or Gram-negative organisms.
If Hartmann (1936) had not included two bacilli in his
study of the action of sodium azide (Table 23) he might
have generalized and concluded that Gram-positive bac-
teria are more resistant than Gram-negative. The re-
sults obtained with the two bacilli proved conclusively
that there was no correlation between Gram staining and
azide sensitivity. We shall, however, present here the
evidence from the literature for and against the corre-
lation.

Diernhofer (1936), in an attempt to find a medium in
which the mastitis streptococcus would outgrow the co-

124

ANTISEPTICS

Ion species, investigated the antiseptic action of many
different compounds. Since this work illustrates the
present problem, we give here (Table 24) the entire list
of compounds he studied. It must be remembered, how-
ever, that Diernhofer's work refers only to streptococci
versus colon-proteus types, and makes no general claim
for differentiation between Gram-positive and Gram-
negative bacteria.

The best-known of the anti-bacterial substances pro-
duced by microorganisms, gramicidine, is so named be-
cause it is supposed to kill only Gram-positive bacteria.
Dubos and Hotchkiss (1941) summarize as follows its
selective action (and that of the allied substance, tyroci-
dine) : "Pneumococci, streptococci, staphylococci, diph-
theria and diphtheroid bacilli, aerobic and anaerobic
sporulating Gram-positive bacilli, have all been found

TABLE 23

Sensitivity of various organisms to sodium azide. (From Hart-
mann, 1936.)

Number

of
strains
tested

Retarding

Inhibiting

Species

concentration
(ppm.)

concentration
(ppm.)

8

Streptococcus agalacticae

350

>500

3

epidemicus

>500

>500

3

dy agalacticae

>500

>500

3

uberis

200

300

3

pyogenes

>500

>500

1

lactis, sucrose-positive

100

400

1

lactis, sucrose-negative

>500

>500

1

lactis, sucrose-negative

100

300

1

Micrococcus sp. from millc

200

>500

1

" " saliva

>500

>500

2

" milk &
manure

400

>500

1

Yeast from mastitis udder

50

50

1

Bacillus subtilis (?)

50

150

1

Bacillus mesentericus (?)

50

150

1

Bacterium coli

50

150

1

Bacterium paracoli

100

150

1

Proteus

50

150

1

Micrococcus, Gram-negative

100

150

ANTISEPTICS 125

TABLE 24

Differences in growth inhibition of streptococci and colon bacteria
by various antiseptics. (From Hartmann, 1936, and Diernhofer, 1933
and 1936.)

Group I: Antiseptics inhibiting streptococci definitely more than
the colon group:

Sodium oleate, 0.3%
Victoria blue, 0.001%
Patent blue, 1.0%
Congo red, 1.0%
Crystal violet, 0.001%
Nile blue, 0.01%
Uranyl acetate, 0.3%
Cusylol, 0.03%
Sodium fluoride, 0.01%

Also: methyl violet, flavine (including its derivatives: scutel-
larein, dimethyl scutellarein and fisetin), quinine, strychnine,
methyl amine, alpha naphthyl amine, pyridium, sodium borate,
thiosulfate, potassium cyanide, ammonium rhodanate, ver-
onal.

Group II: Antiseptics inhibiting the two groups of organisms about
equally:

Hexa methylene tetramine, 0.1%
Phenol, 0.01%
Sulfo salicylic acid, 0.1%
Sodium hippurate, 3.0%
Sodium nitroprusside, 0.5%
Quercitrin, 1.0%
Lead acetate, 0.05%
Barium chloride, 0.3%
Ammonium oxalate, 1.0%.
Chromic acid, 0.05%

Also: formaldehyde, rongolite, chlorophyll, saponin, tannin,
chloral hydrate, potassium ferricyanide, picric acid, diphenyl
amine, anaesthesin, m-nitrophenol, and gamma-dinitro phenol.

Group III: Antiseptics inhibiting the colon group definitely more
than the streptococci:

p-Nitrophenol. 0.05%

«-dinitro phenol, 0.1%

Antipyrin, 1.0% ,

Hydroxyl amine, 0.03%

Hydrazine, 0.03%

Theophyllin sodium salicylate, 1.0%

Sodium azide, 0.02%

Also: pyramidone.

to be susceptible to both gramicidine and tyrocidine. On
the contrary, the following Gram-negative groups, EscTi-
erichia, Klebsiella, Shigella, Salmonella, Hemophilus,
Neisseria, are resistant to gramicidine, but susceptible

126 ANTISEPTICS

to tyrocidine. " To tlie first list can be added a Gram-
positive yeast which was killed by a fairly large dose
of gramicidine and to the second a Gram-negative spore-
former which was not affected by the drug.

These convincing results have been in general con-
firmed by those of Downs (1942). There were, however,
two exceptions : the meningococcus, which was more sen-
sitive to gramicidine than the staphylococcus, and a sen-
sitive avirulent strain of Pasteurella tularensis. These
exceptions limit the validity of the general conclusion
stated above, and, if confirmed, would disprove the con-
ception that susceptibility to gramicidine and the prop-
erty of Gram staining are due to the same mechanism, or
involve the same cell constituents.

For penicillin, the same selectivity is sometimes claim-
ed, but the order of sensitivity as given by Abraham et al.
(1941) disproves this contention. Complete inhibition
is not accomplished by 1,000 ppm. penicillin with Vibrio
cholerae, Salmonella paratyphi and typJiy murium, Kleb-
siella pneumoniae, Escherichia coli, Mycobacterium tu-
berculosis, Pseudomonas aeruginosa and Brucella meli-
tensis. A somewhat higher sensitivity is observed with
Pasteurella pestis, Brucella abortus, Shigella dysenteriae;
250 ppm. inhibit Streptococcus viridans, Proteus vul-
garis, and an anaerobic streptococcus; 100 ppm. suffice
for Diplococcus pneumoniae, and Eberthella typhi, 50
ppm. for Salmonella enteritidis. Then comes quite a gap
in the degree of sensitivity; the next group includes
C orynebacterium diphtheriae, inhibited by 8 ppm, and
Clostridiwm oedematiens by 3 ppm., while 1 ppm. inhib-
its Streptococcus pyogenes. Staphylococcus aureus. Ba-
cillus anthracis and Actinomyces bovis. Still more sen-
sitive are Clostridium Welchii, inhibited by 0.7 ppm.,
and Neisseria gonorrhoeae by 0.5 ppm.

Since, in this list, most of the first-named organisms
are Gram-positive and most of the last-named are Gram-
negative, there appears to be a parallelism between

ANTISEPTICS 127

Gram staining and sensitivity to penicillin (and this has
been quite useful as a guide to medical application).
However, there are striking exceptions, such as equal
sensitivity of Streptococcus viridans and Proteus, or the
extreme sensitivity of the Gram-negative Neisseria gon-
orrhoeae. Typical exceptions, even if few, are enough
to disprove the assumption that Gram staining and sus-
ceptibility to penicillin are fundamentally of the same
origin.

Another frequently made claim is that the basic dyes
inhibit Gram-positive bacteria far more than Gram-neg-
ative. Ingraham (1933) investigated 20 species and ar-
ranged them in the order of their sensitivity to gentian
violet (measured by a quantitative index). In her list
the most sensitive genera were the Gram-positive Bacillus,
Streptococcus and Staphylococcus; but among the most
tolerant organisms were two species of Clostridium,
which are usually considered Gram-positive. However,
since both species are Gram-positive only when young,
and become Gram-negative with age, this observation
can not be considered as an argument against the gen-
eral claim (see, however, the Chapter on Dyes).

Cooper and Mason (1927) studied the differences in
the two outstanding Gram-negative groups, Bacterium
coll and Pseudomonas. They came to the conclusion that
the Pseudomonas group was very sensitive to such
agents as heat or alcohol which produce physico-chemi-
cal changes of the colloidal state or denaturation of the
protoplasm, but rather resistant to substances which re-
act chemically, while the colon group showed the oppo-
site characters. They determined the lowest concentra-
tions inhibiting multiplication for 48 hours. With most
reagents Pseudomonas putida, the most sensitive species
of that genus, was more easily inhibited than Bacterium
coli; but the differences in sensitivity were slight except
with quinol and pyrogallol, of which Bacterium coli could
tolerate more than twice as much as Pseudomonas. With

128 ANTISEPTICS

a few disinfectants, Pseudomonas could withstand strong-
er doses. It could resist 8 times as mucli hydrogen per-
oxide, 6 times as much acrifiavine, 3 times as much qui-
nine and silver nitrate, and twice as much picric acid as
Bacterium coli. These differences are small compared
with those found between Gram-positive and Gram-neg-
ative bacteria.

There is an extensive literature on the addition to cul-
ture media of antiseptics which are meant to do no harm
to a certain desired species, but to suppress most or all
other species. The use of dyes for this purpose is quite
general. In water bacteriology, a soap, sodium ricinol-
eate, is used, and a great number of other media have
been proposed to separate Bacterium coli from other re-
lated species. To give just one instance of the compre-
hensiveness of the literature on that subject, it may be
noted that volume 19 of Milclnvirtschaftliche Forscliun-
gen contains not less than seven papers on selective me-
dia for the isolation of either Streptococcus agalacticae
(mastitidis) or Brucella abortus. It is, of course, be-
yond the scope of this review to discuss all the applica-
tions of the selective action of antiseptics to the isolation
of certain species.

III. DILUTED DISINFECTANTS AS ANTISEPTICS

The mode of action of antiseptics can be studied best
by beginning with the effect of diluted disinfectants.
Their lethal effect at higher concentrations, that is, the
irreversible inactivation of the mechanism of cell divi-
sion, is fairly well understood, but, as was pointed out in
the introduction, the reversible inhibition or retardation
of multiplication very probably is an entirely different
chemical reaction.

To analyze the mechanism of growth inhibition, as was
also explained above, it is not sufficient to know the min-

AXTIHEPTICS

129

imal inhibiting dose of a compound, it is necessary also
to study the complete multiplication curve as atfected
by sub-minimal doses, to measure the length of the lag
period, the multiplication rate and the final *'crop." No
such complete data for any particular compound could
be found in the literature. For the data of Table 25, the
author is indebted to Miss Jean E. Conn, of the Agri-
cultural Experiment Station, Geneva, N. Y. They are
part of a comprehensive study to determine the effect
of acidity on various antiseptics.

It should be pointed out first that the two compounds
which are mentioned in this table and which will be
considered here, formaldehyde and phenol, were hardly
affected by acidity. With formaldehyde there was no
influence whatever; at all acidities included in the test.

TABLE 25

Multiplication of Saccharomyces ellipsoideus in presence of phenol
and formaldehyde. (Unpublished data by Jean E. Conn.)

Phenol

PH

Cone.

(%)

Lag
(hrs.)

Generation
time (hrs.)

Final population
(millions/cc.)

3.1

0.00
0.10
0.12
0.14

2-3
2-3
2-3
3-6

1.20

1.88
3.18
3.48

60
30
30
14

5.3

0.00
0.10

2-3
60

1.29
6.4

40
17

Formaldehyde

pH

Cone.

(%)

Lag
(hrs.)

Generation
time (hrs.)

Final population
(millions/cc.)

3.1

0.000
0.004
0.006

2-3

12
60

1.20
2.58
3.38

60
40
30

5.3

0.000
0.004
0.006

2-3
15-18
50-55

1.29
3.15
3.66

40
40
46

130 ANTISEPTICS

slow growth was observed with 0.006% formaldehyde,
but none with 0.008%. The phenol efficiency was slight-
ly altered by acidity ; at pH 5.3, complete inhibition was
obtained by 0.12%, while at pH 3.1, slow growth was
observed with 0.14%.

The effect of phenol consisted largely in a decrease of
the rate of multiplication (Fig. 24). There was no in-
crease in the lag phase. The generation times, computed
from plate counts, are represented in Table 25. Direct
microscopic counts of dead cells were also made after
mixing 1 cc. of the phenol-treated cultures with 1 cc. of
of 0.01% methylene blue solution. The proportion of
dead cells never exceeded 5%. This observation per-
mits the conclusion that the increased generation time
was not the result of excessive death in a population
which otherwise was multiplying normally, but was due
to a lower average growth rate. The final number of
cells capable of growing with or without phenol was de-
termined only roughly, but unquestionably it was small-
er in the presence of the disinfectant.

Formaldehyde produced the same symptoms of de-
creased rate of multiplication and decreased maximal
jDopulation, but in addition it caused a quite extended
lag (Table 25 and Figure 24).

As far as the observations of the writer go, formal-
dehyde is the only strong disinfectant which causes such
a marked prolongation of the lag period. Mercurochrome
also increases the lag, but to a much lesser extent. The
inhibition of growth by formaldehyde is so prolonged
that in determining the phenol coefficient of this disin-
fectant, the usual observation time of 48 hours gives
wrong results and nnist be extended to 4 days.

These two diluted disinfectants are alike in one re-
spect: as long as the dose is sufficient to prevent growth,
they kill the cells slowly but continuously. With weaker
doses, a slow growth sets in, but many cells die before
multiplication starts. The rate of multiplication de-

ANTISEPTICS

131

6

U
(U

a

7 —

cJ

_..--

1

"

in

O

1/1

o

V

/

/'^

/

£

C

O

^

. Q

Z

/"

,^

y

^

^^'--

/

3

« 12 IB 2<^--^ *?- ' *" ^ ^'^ "O""*

«6

Fig. 24. Multiplication curves of Saccharomi/ces ellipsoideus in
presence of various concentrations of phenol and formaldehyde. (Orig-
inal data.)

creases as the concentration of the disinfectant increases,
and becomes zero at the inhibitive dose.

In addition to the decreased multiphcation rate, for-
maldehyde causes a pronounced increase in the lag period
which phenol does not show. Thus three types of inter-
ference are possible ; extension of lag with normal growth
rate, decrease of growth rate with normal lag, and de-
crease of growth rate plus extension of lag. All three
types will be shown to actually occur.

Another question arises when diluted bactericides are
used as antiseptics, namely, that of the ratio of inhibi-
tory to bactericidal concentration. On this point quite
extensive data have been presented by Birkhaug (1933),
and the ratio was found to fluctuate greatly. Table 26
gives a few of the reported data. AYith the last three
disinfectants, the inhibition of growth (without killing)
of Bacterium coli requires 30 to 40 times as high a con-
centration as the corresponding inhibition of Staphylo-
coccus. But the dose necessary to kill Bacterium coli is
only 3 to 4 times as high as that for Staphylococcus. If
irreversible sterility (or death) were due to the same
reaction as reversible sterility (or antisepsis), death

132

ANTISEPTICS

should occur when antisepsis has reached a certain stage,
e.g., when 99% of certain molecules are inactivated. The
reaction could reach the same stage (death) at the same
time (10 minutes) only if Bacterium coli were exposed
to a concentration 40 times as high as that for Staphylo-
coccus. However, the experiment shows that the concen-
tration for death need be only 4 times as high for Bac-
terium coli. Therefore, disinfection has a different con-
centration coefificient than antisepsis, and that means a
ditferent reaction.

The above-mentioned experiments by Jean Conn have
shown that the inhibiting concentration for yeast is 0.16%
with phenol and 0.008% with formaldehyde. The killing
concentrations are approximately 1.0% and 2.5% for 10
minute death times. The ratios of the killing to the in-
hibiting concentrations are thus: 1:0.16=6.25 and

TABLE 26

Inhibitory and lethal doses of mercury compounds.
(From Birkhaug, 1933.)

Dilution

needed

To kill in
10 minutes

To inhibit
growth

To kill in
10 minutes

To inhibit
growth

Staph, aureus

Bact. coli

Phenol

Mercury chloride
Mercurochrome, 2%
Metaphen
Merthiolate
Phenyl mercuric
nitrate

1: 85

16,000

160

140,000

120,000

192,000

1: 800

18,000

18,000

18,000,000

26,000,000

120,000.000

1: 75

10,000

180

32,000

32,000

48.000

1: 800

34,000

3,200

400,000

860,000

4,000,000

Ratio of lethal to

inhibitory dose

Phenol

Mercury chloride
Mercurochrome, 2%
Metaphen
Merthiolate
Phenyl mercuric
nitrate

9:1
1.1:1
110:1
130:1
216:1

630:1

11:1
3.4:1
18:1
12.5:1
27:1

83:1

ANTIi^EPTICS

2.5:0.008=320 respectively. The conclusion, here again,
is that the reaction inhibiting multiplication seems to be
different from that causing death.

IV. EFFECT OF SIZE OF INOCULUM

Many statements can be found in the literature to the
effect that a certain concentration of an antiseptic will
inhibit bacterial multiplication if the inoculum is small,
while growth will take place if the inoculum is large ( see
references in papers by MacLeod and Mirick, 1942, and by
Lee et al., 1943), This may result from different causes.
The most obvious is the presence of organic matter in the
inoculum. Ordinarily, the inoculum consists of a small
volume of culture which contains some organic matter
besides the bacteria. Substances such as chlorine, io-
dine, or permanganate, react promptly with organic mat-
ter, and their concentration is thereby decreased. A
small inoculum will have little influence, but a large inoc-
ulum may reduce the concentration of the antiseptic be-
low the inhibiting dose. If the inoculum consists of
washed cells, the effect of its size on the concentration
of the antiseptic will be much smaller.

If the inoculum contains large colloid particles, an-
tiseptics of the type of aniline dyes or hexyl resorcinol
may be adsorbed, and the decrease of available antiseptic
will be larger if the inoculum is larger.

Quite generally, if bacteria produce no visible growth
in a medium containing an antiseptic, the cause of un-
observable multiplication need not always be the same.
It may be either the death of all the cells, or the perma-
nent, though reversible inhibition of multiplication. But
two other causes must also be considered, namely, too
early a cessation of observation which did not give the
bacteria enough time to produce visible turbidity, or too
small an increase in numbers of bacteria to produce visi-

134 ANTISEPTICS

ble turbidity. Each of these causes may, under certain
conditions be modified by the size of the inoculum.

When antiseptics cause death, they act very slowly,
and with many antiseptics, the lethal action ceases after
some time. In the case of dyes, the survivors can grad-
ually modify the antiseptic so that multiplication be-
comes possible. If 99% of the cells die before growth can
start, all the cells of an inoculum of 50 cells may be kill-
ed, while with a large inoculum, a sufficient number will
survive to start growth again. In this case, the bacteria
themselves modify the antiseptic, namely, the dye, and
make it less toxic (for details, see next chapter).

With volatile substances like alcohol, chloroform, thy-
mol, formaldehyde, etc., the concentration may gradually
decrease by evaporation of the antiseptic from cultures
closed by a cotton plug. With a small inoculum, all cells
may be dead before the decrease in antiseptic permits
multiplication, while a large inoculum has far greater
chances of having some survivors.

Erroneous conclusions as to the effect of the size of the
inoculum can further result from too short an observa-
tion time. The usual inoculation procedure consists in
transferring a loopful (0.01 cc.) from a full-grown cul-
ture to a test tube containing the medium with the anti-
septic. This amounts to an initial number of about a
million cells per cc. of culture. With medium-sized bacte-
ria, turbidity becomes noticeable with about 5 million
cells per cc. If the bacteria require 2 hours to double
their number, they need only 3 generations or 6 hours to
produce visible turbidity. To this must be added the
lag period which would not be more than 3 hours with
such a heavy inoculum. If the antiseptic decreases the
rate of multiplication to one-fifth, bacteria would require
10 hours to double their number, and the total time to
produce visible turbidity would be 30-|-3 = 33 hours.
This large inoculum would thus produce turbidity in 9
hours without antiseptic, and in 33 hours with antiseptic.

ANTISEPTICS 135

If another set of cultures is made Avith only 1,000 cells
per cc, instead of 1 million, we must add to the above
figures the time needed by the bacteria to multiply from
1,000 to 1,000,000. This corresponds to 10 generations,
which require 20 hours without antiseptic, but 100 hours
Mdth antiseptic. We must further consider that a smaller
inoculum demands a longer lag period, approximately 12
hours, in our case, instead of 3 hours. Thus, the control
culture without antiseptic would produce visible turbidity
in 6+20-|-12 = 38 hours while the culture with antiseptic
would need 30+100-f 12 = 142 hours = 6 days. With a
still smaller inoculum of only 1 cell per cc, the lag period
would be about 24 hours, and the total time to produce
turbidity would be 3 days without antiseptic and 10 to 11
days with antiseptic. Since experiments are often dis-
continued if no turbidity is noticeable after one week, an
antiseptic may be believed to suppress multiplication
completely, w^hile bacteria have been increasing contin-
uously, although at a very low rate. A decrease in the
rate of multiplication has been observed with most anti-
septics, e.g., with sulfanilamide, benzoic acid, SOo, peni-
cillin, etc. Not all of them, how^ever, cause also an in-
creased lag period.

That an increased lag can give the appearance of com-
plete inhibition of growth, is obvious See Fig. 25).
Cruess and Richert (1929) found that in grape juice ad-
justed to pH 6.0, benzoate of soda retarded yeast devel-
opment greatly. With 0.6%, growth began after 12 days,
while with 0.8% growth started only on the 25th day.

An interesting complication is encountered in the re-
sults of Lee, Epstein and Foley (1943) who studied the ef-
fect of urea on the development of Bacterium coli (Table
27). Growth was recorded by turbidity measurements.
(Eoughly, the turbidity is linearly related to the popula-
tion.) After 21 hours, the population was higher in the
cultures containing urea than in those without it, espe-
cially wdth the smaller inocula. This indicates an in-

136

ANTISEPTICS

12 3 4 5 6 DAYS eoi2J«58»a OAYS q

Fig. 25. Effect of the amount of inoculum on the time required for
the development of visible turbidity in a bacterial culture, when the
antiseptic prolongs the lag period (left), and when it retards the
growth rate (right). The broken lines refer to controls, the solid
lines to cultures with antiseptic. Abscissa: time; ordinate: logarithm
of the number of bacteria per cc.

creased growth rate. But the 5-hour measurements
show a decrease of growth caused by urea. The two
observations can be explained only by the assumption
that urea prolongs the lag phase, but increases the growth
rate. The example indicates how important it is to con-
sider these two phenomena separately.

V. THE DUAL ACTION OF THE DYES

Dyes have been employed in the treatment of wounds
as early as 1890, and since 1900 many attempts have been
made to develop dye derivatives for chemotherapeutic
purposes. Several such compounds have now been in
use for some time, but, in general, they have proved more
eifective against trypanosomes than against bacteria.
The outstanding fruits of this search for therapeutic
agents are rivanol and germanin (Bayer 205).

Like many other antiseptics, dyes in high concentra-
tions act as disinfectants, but high concentrations are
rarely used. In the customarj^ w^eak solutions, their
effect differs from that of most antiseptics. The bacteria

AXTI^'^EPTICS

1:57

TABLE 27

Effect of size of inoculum on growth of Bacterium coli (shown by
turbidity measurements), in cultures containing urea. A value of
30 indicates 650,000,000 cells per cc. (From Lee, Epstein, and Foley,
1943.)

Number of cells per
mm^ at start

65

650

6,500

60,000

Observation time
(hours)

5

21

5 21

5 21

5 21

Cone, of urea

(%)

Turbidity

0

0.2

0.5

1.0
2.0

0
0
0
0
0

14
24
26
28
26

3 19
2 25
2 28
0 33
0 25

18 36
27 26
16 32
11 40
9 24

28 32

28 28
34 29

29 36
28 35

are inactivated only for a time ; after a while they begin
to grow in the presence of the dye, and then multiply at a
normal rate, showing no trace of injury. It seems as if
the dye merely prolongs the lag phase without producing
any other effect. Churchman (1912) coined the term
''bacteriostasis" for this phenomenon.

The difference between dye bacteriostasis and the de-
crease of growth rate caused by most other antiseptics
is shown in Figure 26. The graph on the left side shows
the effect of rivanol (an acridine dye) on the oxygen
absorption of Bacterium coli. Since the oxygen absorp-
tion is proportional to the bacterial concentration, the
curves of Fig. 26 are really growth curves. The higher
concentrations cause a longer lag, but they do not af-
fect growth rates, as is indicated by the fact that the
growth curves remain parallel. The graph of the right
side represents the action of sulfa-thiazole. This com-
pound lowers the growth rate ; consequently the slope of
the growth curves decreases when the concentration in-
creases, and the curves spread out fan-shaped. All curves

138 ANTISEPTICS

seem to have started at the same time; there was no
increase of the lag period.

This very unusual effect of the dyes was explained by
Dubos (1929) as due to a change of the oxidation po-
tential of the medium by the dye. He showed that a
Pneumococcus R II and a Streptococcus liemolyticus
were inhibited only by those dyes whose oxidation poten-
tial was higher than that corresponding to an rH of 12.5
while they grew in all dyes with a lower potential.

Ingraham (1933) verified Dubos' tentative explana-
tion in an extensive study of the action of gentian vio-
let. The relationship that she found between the length
of the lag phase and the logarithm of the cell concentra-
,tion is rectilinear and can be expressed by

Lagi = Lago + k log No/Ni

where N, represents the larger and Ni the smaller inocu-
lum, and Lag2 and Lagi the corresponding lag phases.
The proportionality constant h depends upon the species
as well as on the medium. She furthermore showed that
a large inoculum may be capable of adjusting the poten-
tial to the optimum of cell growth so that the bacteria can
multiply again, while a small number of cells is not able to
accomplish this change.

Hoffmann and Rahn (1944) separated the bactericidal
from the bacteriostatic action. In concentrations over 4
ppm. of crystal violet (the purest form of gentian violet
available), in which Streptococcus lactis was killed, the
order of death was logarithmic, and the deathrate con-
stant was practically proportional to the dye concentra-
tion K = 0.037 c, where c is the concentration in ppm.
(See Table 28). Death is probably due to the reaction
of the dye with some cell constituent essential for a vital
function such as cell division, as Stearn and Stearn
(1924) had claimed. It is not due to inactivation of en-
zymes, as Hoffmann's (1943) experiments proved.

ANTISEPTICS

139

TABLE 28

Effect of crystal violet on Streptococcus lactis in broth cultures.
(Original data.)

.

Deathrate

constant

Crystal Vioiet

K

Beginning of

growth

Dilution

ppm.

Calculated

Measured
0.08 to 0.16

1:1,500,000

0.67

0.025

after 31 hours

1,000,000

1

0.037

0.09 to 0.17

after 3 days

750,000

1.33

0.050

0.08 to 0.11

no growth in 3 days

500,000

2

0.074

0.16 to 0.19

"

400,000

2.5

0.093

0.12 to 0.20

no growth in 2 days

250,000

4

0.I4S

0.11 to 0.30

200,000

5

0.185

0.19 to 0.21

100,000

10

0.370

0.34 to 0.37

50,000

20

0.740

0.67 to 0.80

10,000

100

3.700

3.13 to 5.36

9 6 7 8 hOMrs 3 4 5 6.7 8 hours

Fig. 26. Different types of antiseptic action. Left: multiplication
curves of Bad. coH obtained with various concentrations of rivanol,
a dye which increases the lag period but does not affect the growth
rate; the curves are parallel. Right: multiplication curves of Bact. coli
obtained with various concentrations of sulfathiazole, a dye which
retards the growth rate, but does not affect the lag; the curves spread
out fan-shaped. In the two graphs the oxygen consumption is con-
sidered a measure of the number of organisms. Abscissa: time in
hours; ordinate: oxygen consumption, on logarithmic scale. (From data
of Hirsch, 1942.)

140 ANTISEPTICS

In solutions containing less than 4 ppm. of crystal vio-
let, the death rate of S. lactis did not change with the
dye concentration, but remained nearly constant. The
bacteria were still dying at these low concentrations, and
the order of death was logarithmic for a while, but later
the death rate decreased. If the inoculum had been suf-
ficiently large, the surviving bacteria Avould have begun
to grow in the presence of the dye which a short time
ago had killed the majority of the inoculated cells, and
they would have multiplied as rapidly as if no dye were
present.

The fact that the death rate is independent of the
concentration in the range below 4 ppm. and that bacte-
ria multiply later without any indication of injury fits
well into the explanation given by Dubos. In dilute dye
solutions bacteria will die largely on account of the ab-
normal reduction potential which is the same in high and
low concentrations of the dye. Of course, some cells are
also dying from the direct reaction of some of their
constituents with crystal violet, but in these great dilu-
tions, the rate of death from that cause will be very low
and become negligible in comparison with the rate of
death from the abnormal potential.

Bacteria can cliange the potential of this medium as
they change that of any other medium in which they
grow. The change is broug'ht about by their metabolism,
largely by their reducing power. If the inoculum is large,
the change is rapid; if it is small, the change is slow.
With a verj^ small inoculum, all cells may be dead be-
fore the potential is sufficiently adjusted to permit
growth. Once the jDotential is adjusted, multiplication
is not impeded, and the rate of multiplication is not de-
creased by the dye. Hoffmann and Rahn also showed that
aeration lengthens the time of ''recovery" of the cul-
ture because it keeps the oxygen concentration of the
medium high, and counteracts the reduction by bacteria
while removal of oxygen hastens it. The marked effect of

AXTISEPTICH

141

pH, which controls the reduction potential, on the length
of las: is shown in Figure 27.

Fig. 27. Survivor curves ot strepiococcus lactis treated witli vari-
ous concentrations of crystal violet. (From data of Hoffmann, 1943.)

The assumption of a dual role of the dyes is supported
by their effect on cells of different ages. The stronger
solutions of crystal violet kill young cells more rapidly
than old ones, as in fact, all disinfectants do^ In di-
lute solutions, multiplication sets in sooner when the

1. See p. 42.

142 ANTISEPTICS

cells are very young because young cells have a much
more active metabolism than old ones, and therefore can
adjust the potential more readily than old, resting cells.

That an unfavorable oxidation-reduction potential re-
tards or prevents multiplication is common knowledge.
That it kills the bacteria fairly readily represents a new
experience. Ingraham (1933) made the statement that
bacteria are not killed by crystal violet, but Hoffmann's
numerous data (1943) show that even with dilute dyes,
99% and more of the cells are killed before multiplica-
tion begins. That these cells are really dead is evidenced
by the fact that activated charcoal, which removes all
crystal violet from the culture in less than a minute, did
not restore in them the faculty of reproduction.

If antisepsis by dyes is due to the establishment of a
redox potential too high for multiplication, the species
with the strongest reducing power, the Clostridia, should
be the most tolerant to dyes, with the colon group as a
close second, while the strict aerobes should be more sen-
sitive than the streptococci. Ingraham has arranged 20
species in order of increasing sensitivity, and, with a few
exceptions, the observed order agrees with that which
general conceptions of ''reducing power" in these spe-
cies would predict. Clostridia are, in fact, the most tol-
erant species, while the aerobic spore-formers and the
aerobic yeasts Torula and Monilia are the most sensitive
ones. The observation of Slanetz and Eettger (1933)
that the fusiform bacteria can tolerate gentian violet in
concentrations of 500 ppm. (1:2,000 dilution) fits into
the picture because this group consists of strict anae-
robes. The highly selective action of this and other dyes
is probably not connected with the Gram staining reac-
tion, as is commonly thought, but depends upon the re-
ducing power of the species. Because of this selectivity,
some dyes have been added to special culture media to
suppress certain groups while permitting others to grow.

AXTh'^EFTICS

143

The quantitative relations between the length of the lag
period and cell concentration, established by Ingraham
(p. 138) were verified by Hoffmann (see Fig. 28). This
is difficult to explain. If one culture were inoculated with
1000 times as many bacteria as the other, metabolism
would be 1000 times as intense and adjustment of po-
tential 1000 times as fast, so that the favorable poten-
tial should be reached in one-thousandth of the time. The
length of the lag period would be one-thousandth of Lagi,
while in reality the relation is

Lagxooo = Lag, + k log 1/1000
= Lagi — 3 k,
k being a proportionality constant. No explanation of
this phenomenon has been offered.

The antiseptic power of dyes is greatly affected by the
acidity of the medium (see Fig. 29). Acid dyes {e.g.
acid fuchsin, eosin) — which in general are rather weak

Fig. 28. Effect of the amount of inoculum on tlie lengtli of the
lag phase of Bacterium coli treated with crystal violet. Abscissa: lag
phase; ordinate: concentration of inoculum in number of bacteria
per cc, on logarithmic scale. (From data of Hoffmann, 1943.)

144

ANTISEPTICS

antiseptics — act most efficiently in acid media, while basic
dyes {e.g., gentian violet, basic fucbsin, all flavine dyes)
are more bacteriostatic in alkaline media (see Table 29).
Hoffmann (1943) found that with gentian violet, only the
bacteriostatic effect changes with pH while the bacteri-
cidal effect is not altered appreciably. As the reduc-
tion potential of dyes and of most culture media is great-
ly affected by pH, it is not surprising that their bacterio-
static action changes correspondingly. Hoffmann further-
more showed that the effect of acidity upon bacteriostasis
had nothing to do with the electrolytic dissociation of the
dye.

While the investigation of the detailed antiseptic
mechanism of the dyes has been worked out almost en-
tirel}^ with crystal violet, the principles apply to many
other dyes. As already mentioned, Dubos (1929), com-
paring the effect of 15 dyes, found that the oxidized indo-

Fig. 29. Effect of acidity on the length of the lag phase of Strep-
tococcus lactis treated with various concentrations of crystal violet.
(From data of Hoffmann, 1943.)

ANTISEPTICS

145

TABLE 29

Bacteriostasis of Bacterium coli by various dyes. (From Steam and
Stearn, 1926.) + indicates growth; — indicates no growth.

Basic Dyes

PH

4.95 1 5.28

6.23 1 6.46

7.16

7.73

Dilution

Gentian Violet

1: 10,000

20,000

30,000

70,000

100,000

200,000

+
+
+
+
+

+
+
+

+
+

+ 1 1 1 1 1

Brilliant Green

50,000

75,000
150,000
300,000
600,000

+
+
+
+

+
+
+

+

*

+
+
+
+

Methyl Violet

10,000
20,000
30,000
70,000

+
+
+
+

+
+
+

+

+

Acid Dyes

pH

4.95

5.28 6.23

7.16

7.73

Dilution

Acid Fuchsin

25
50

75
100

—

+

+
+
+

+
+
+
+

Eosin

25

50

75

100

+

+

+
+

+
+
+

+
+

+

Acid Violet

25

50

75

100

+

+
+

+
+
+
+

♦The color of brilliant green fades out at pH above 7.0

146 AXTISEPTICS

phenols and methylene blue were bacteriostatic for Pneu-
7nococcus and most hemolytic streptococci, while the in-
digoes, malachite green and the indophenols were not
toxic. In the reduced state, methylene blue and the in-
dophenols were not toxic either.

VI. MODE OF ACTION OF THE SULFONAMIDES

The antiseptic properties of sulfanilamide and its de-
rivatives have probably been studied more intensely than
those of all other antiseptics combined. The review by
Henry (1943) lists 291 references. The sulfonamides are
not strong disinfectants, but owe their fame to their
ability to prevent, in concentrations which do no great
damage to blood or tissues, multiplication of many spe-
cies of bacteria.

The sulfonamides differ from all other antiseptics in
that they do not affect the bacteria at once, but require
a measurable amount of time before bacterial gro^vth is
retarded. The delay varies greatly, usually from 2 to 6
hours, according to Henri's compilation of data from
seven authors. With all other antiseptics, the degree of
growth retardation is uniform as long as the culture
grows, or if the rate of multiplication occasionally va-
ries, it may increase, as with crystal violet, but it never
is normal at the start and decreases later. The writer
has checked all his plate count and turbidity rec-
ords of the early stages of antisepsis by phenol, formal-
dehyde, benzoic and sulfurous acid, and in every case
multiplication was promptly retarded even by small doses.
A picture like that of Figure 30, which was obtained by
Muir, Shamleffer and Jones (1942) in their study of the
effect of sulfathiazole on Bact. enteritidis is characteris-
tic only of sulfonamides. These authors observed that
Bact. enteritidis treated with sulfathiazole multiplies for
a short time quite normally, but after about 3 genera-

AXTISEPTIC8 147

CONTROL -NO DRUG

Fig. 30. Multiplication curves of Salmonella enteritidis with and
without sulfathiazole. Abscissa: time in hours; ordinate: logarithm
of the number of organisms per cc. (From Muir, Shamleffer and
Jones, 1942.)

tions the generation time in broth increases from 17 to 90
minutes with 50 piDm. thiazole and to 120 minutes with
100 ppm. In a synthetic medium, 6 ppm. is sufficient to
stop growth completely after multiplication has taken
place quite normally for 10 hours. The type of curve
suggests a deficiency of a necessary cell construction
material in the medium.

This delay is not caused by a very slow penetration of
the drug into the cells, as is proved by the fact that
the luminescence of three species of photogenic bacteria
was inhibited without delay as soon as the cells came
in contact with sulfanilamide (Johnson and Moore, 1941).

The assumption of a slow penetration is also contra-
dicted by the following studies of Kohn and Harris
(1941a) with resting cells: They inoculated Bacterium
coli into two tubes of culture medium held at 5°, one of
which contained a retarding dose of sulfonamide. (At 5°,
Bacterium coli does not grow.) After 3 hours, an equal
amount of sulfonamide was added to the other tube, and
both were incubated at 30°. The delay of the growth re-

148

ANTISEPTICS

tardatioii was the same; the contact of the resting cells
with the drug for 3 hours had been without any influence.
In another experiment, the bacteria were held without
food in presence of sulfonamides for 3 hours at 37°, When
the food was added, the culture multiplied normally for
a while, thus showing that the delay of the drug effect
had not been changed by a 3-hour contact of the culture
with the drug, because there had been no growth. Both
experiments were made with four different sulfonamides.
The authors concluded that ''the interaction between the
sulfonamide and the bacterium, therefore, depends upon
some reaction associated intimately with growth".

Another noteworthy difference between sulfonamides
and other antiseptics has been pointed out by Kohn and
Harris (I.e.). It concerns the relation between con-
centration and retardation of growth. Very much more
sulfonamide is required to produce strong retardation
than slight retardation. Their main data are given in
Table 30, together with results obtained by the writer

TABLE 30

Retardation of growth of Bacterium colt and of yeast by various
antiseptics.

Antiseptic

Concentration required to reduce
the growth rate

30%

85%

Ratio of the

two

concentrations

Bacterium coli (Kohn and Harris, 1941a)

-6

-6

Sulfanilamide

500 X 10 M

5000 X 10 M

10

Sulfapyridine

20

2500

125

Sulfathiazole

4.2 "

500

120

Sulfadiazine

0.84 "

1000

1200

Yeast (Rahn, unpublished data)

Phenol

Formaldehyde
Benzoic acid

0.10%

0.003%

0.015%

0.17%

0.007%

0.040%

1.7
2.3

2.7

ANTISEPTICS 149

with other antiseptics. The table shows the concentra-
tions necessary to decrease the growth rate 30% and 85%.
The difference betw^een them is significant. It is usually
explained by the assum^Dtion that a low concentration of
sulfonamide retards growth slightly by attacking only one
locus of the cell constituent involved in the reaction.
Greater retardation can be brought about by an attack
on other loci, or possibly by some secondary reaction be-
tween the drug and the cell constituents, which would
cause a shift of equilibrium, and require a much higher
concentration of the drug. Several authors have re-
ported indications that other loci are attacked by larger
doses. However, this implies that the first retardation
is not increased by higher concentrations, and this impli-
cation again would make the action of sulfonamides dif-
ferent from that of all other antiseptics.

The frequent!}^ encountered statement, that retarda-
tion of multiplication is caused by a reaction of the sul-
fonamides with catabolic enz;ymies of the cell, has very
little experimental support. Many investigators have
shown that large doses will retard or inhibit enzyme ac-
tion, but the same is true with large doses of sodium chlo-
ride. The inhibition of enzyme action by large doses of
any compound can certainly not be considered proof that
growth retardation by small doses is due to decreased
enzyme action. In order to prove that retardation of
growth is caused by enzyme inactivation, it is necessary
to demonstrate that enzymes are affected by doses caus-
ing a slight retardation of growth, say 20 to 30%. As
early as 1937, Mellon and Bombas measured b}^ the
Thunberg technique the effect of sulfanilamide on the
dehydrogenase of Pneumococcus Type I, and found that
even a concentration as high as 0.17% did not decrease
enzyme activity whereas 0.01% decidedly retarded growth.
Barron and Jacobs (1937) verified this observation for
other species. Chu and Hastings (1938) observed small
decreases (occasionally, however, as high as 50%) in the

150

ANTISEPTICS

oxygen uptake of pneumococcus by doses which inhibited
growth almost completely. Kohn and Harris (1941)
found that a notable decrease in the respiration of Bact.
coll occurred only with very large doses of sulfonamides.
Amounts which produced a 30% retardation of growth did
not atfect the oxygen uptake.

The results of Dorfman et al. (1940, 1941) with dysen-
tery bacteria deal almost entirely with the interaction
between nicotinic acid and some sulfonamides, which has
nothing to do with the fundamental sulfonamide effect,
because sulfanilamide does not show this reaction. The
few data which can be applied to the main problem show
that a much larger dose is needed to retard respiration
than to retard growth. Growth was retarded slightly
by 1 mg. in 100 cc, it was completely inhibited by 30
mg,, but 120 mg. were needed to produce a 15% de-
crease of the oxygen uptake.

This agrees well with the extensive study by Hirsch
(1942). Table 31 shows some of his data. While 0.15
ppm. of sulfathiazole reduces the multiplication rate of
Bacteriu coU 75%, 255 ppm., or 1700 times the growth-

TABLE 31
Effect of sulfa-thiazole on Bacterium coli. (Fi'om Hirsch, 1942.)

Concentration:

Respiration:

Generation time

(mg. per liter)

(mm.3 0^ in 4 hrs.)

(in minutes)

255.—

130.1

no growth

102.—

136.9

"

40.75

139.3

"

16.4

137.8

"

2.62

131.5

"

0.15

128

0.075

....

83

0.0375

60

0.0189

....

53

0.00945

...•

35

Control

135.1

31

Estimated number

of cells per cc.

200,000,000

4,000,000

AXTISEPTICS 151

retarding dose, does not decrease the oxygen uptake at
all.

Greig and Hoogerlieide (1941), working with sulfan-
ilamide, came to the conclusion that ''germicides in bac-
teriostatic concentrations have no effect on metabolic
rates (oxygen uptake) of bacteria, but inhibit multipli-
cation ' '.

MacLeod reported (1939) that sulfapyridine in doses of
125 ppm. retards the action of the dehydrogenase of pneu-
mococci for glycerol, lactate and pyruvate, but not that
of glucose-dehydrogenase. No records are given about
the tolerance of the bacteria involved, but it seems safe
to assume from a preceding paper by MacLeod and
Daddi (1939) that 60 ppm. caused complete inhibition
of the original strain.

The only data showing a real correlation between re-
tardation of respiration and retardation of growth are
those of Sevag and Shelburne with streptococci (1942a)
and pneumococci (1942b). The parallelism is very strik-
ing in some of the experiments with streptococci in which
the sulfonamides reduced both the oxygen consumption
and the rate of multiplication by the following amounts :
Decrease in Decrease in

oxygen consumption rate of multiplication

59% 59%

62% 66%

68% 65%

64% 69%

64% 64%

An equally close agreement was obtained with anaero-
bic cultures where the formation of acid was measured.
These results contradict all the others previously men-
tioned, probably because they represent abnormal growth
conditions. Sevag and Shelburne used inocula of nearly
a billion cells per cc, that is, about the maximal popula-
tion obtainable in broth. They also used very large

152 ANTISEPTICS

amounts of sulfanilamide. While the concentration of the
latter needed to retard by 85% the growth of B.coli is
0.005M, they use 8 times .^s much (0.04M or 0.69%) for
the much more sensitive streptococcus.

Wyss, Strandskov and Schmelkes (1942), following
the Sevag technique, obtained a decrease of 20-28% in
the oxygen uptake of Strep, pyogenes with 0.04 M sul-
fanilamide, but the same or larger decreases were ob-
tained with the ortho and meta derivatives which are
chemotherapeutically inactive. The results with Bad.
coli and Staph, aureus were similar. If these compounds
retard respiration without producing the true sulfona-
mide action, we cannot very well explain the sulfonamide
action as the result of retarded respiration. The same
conclusion must be reached from the findings of Sevag^
that ''approximately 65% inhibition of aerobic respira-
tion or approximately 45% inhibition of anaerobic respir-
ation results in (or accompanies) complete bacterio-
stasis."

This checks with Ely's result (1939) that the respira-
tion of Bad. coli is decreased about 50% by 0.8% sulfan-
ilamide, "with little or no killing of bacteria". In other
words, multiplication ceases completely when the cells
are capable of producing 35 to 55% of their energy by
respiration or fermentation.

However, bacteria do not cease growing under such
circumstances. Foter and Eahn (1936) have shown that
at low temperatures, streptococci and lactobacilli need
only half as much energy for doubling their number as
at the optimum temperature, but are capable of multi-
plying slowly and continuously with only 5% of the
energy output observed at the optimal temperature. The
complete inhibition of multiplication of cells which still
show respiration or fermentation can be due only to a

1. Quoted from Henry (1943), p. 226.

ANTISEPTICS

153

reaction of the drug with some cell constituents which
are more directly related to growth or cell division.

The first attempt to explain sulfonamide action on
bacteria as a direct effect on the growth mechanism is
probabl}^ that by Mcintosh and Whitby (1939) who stated :
''Sulfonamide drugs are not simple germicides. They
probably act by neutralization of some metabolic func-
tion or enzymatic activity". This idea was elaborated
by Wood (1940) who observed that the effect of sulfan-
ilamide is counteracted by p-amino benzoic acid. This
compound, which is found in small amounts in peptone
and meat extract, is necessary for the formation of cer-
tain indispensable cell constituents. Many bacteria can
synthesize it. Those species w^hich produce it freely,
e.g., bacteria of the colon grouj), are able to tolerate
large doses of sulfonamides. Resistant variants of sen-
sitive species have been shown to produce greater amounts
of p-amino benzoic acid than their parent strains (Landy
and assoc, 1943).

All sulfonamides are "antagonized" by p-amino ben-
zoic acid, and this antagonism can be explained in sev-
eral ways : 1. The sulfonamide may combine with p-amino
benzoic acid and make it unassimilable. (That explan-
ation, however, has been disproved.) 2. The sulfonamide
may interfere with that part of the cell which synthe-
sizes p-amino benzoic acid. 3. On account of the great
similarity of the two compounds, the synthetic mechan-
ism of the cell may use sulfanilamide instead of p-ummo
benzoic acid and build it into cell structures which then

COOH

SO2NH2

H

/\

H

H

/\

H

H

\y

H

H

\y

H

NH2

NH.

p-amin(

) ben

soic acid

sulj

'anila

mide

154 AXTISEPTIC8

fail to function because of the abnormal sulfone side-
chain. This would cause a retardation of growth propor-
tional to the number of abnormal side-chains per cell,
and eventually a complete cessation of growth. The
compounds thus incapacitated may include enzymes, but
the decrease of enzymatic activity would be the result
rather than the cause of abnormal growth.

The delayed action of sulfonamides can be explained
by postulating a certain store of ^^-amino benzoic acid
in the cell. The sulfonamide cannot retard growth until
this store is exhausted. The results of Fig. 30 agree
well with the assumption that the new cells are accumu-
lating increasing numbers of inactive, or one might say,
defunct molecules.

A very interesting parallel to the above case has been
obtained with the sulfone derivative of nicotinic acid.
Mcllwain (1940) tested the action of pyridine sulfonic
acid amide on bacteria which need nicotinic acid amide.
The compound retarded the growth of staphylococci, and
this effect could be counteracted by nicotinic acid amide,
just as that of sulfonamides is counteracted by p-ammo
benzoic acid. In a later publication (1942) he gave as
further example of this phenomenon the counteraction of
a-amino sulfonic acids by a-amino carboxylic acids in
Proteus, and of acriflavine by amino acids in Bad. coli.
The above explanations are often presented in various
forms. Johnson (1942) considered all of them improb-
able because one molecule of p-amino benzoic acid is
suJBficient to neutralize the effect of 23,000 molecules of
sulfanilamide on luminescent bacteria.

Fox and Rose (1942) explained this ratio as resulting
from the ionization of sulfonamides, the small disso-
ciated fraction only being effective. They tested the
minimum growth-retarding dose of the drug at pH 7, and
the minimum dose of p-sHinmo benzoic acid which would
neutralize the effect. The ratio obtained varied greatly
from one to the other of the four sulfonamides tested, aa

AXTISEPTICH 155

may be seen in tlie following table, but when tlie amouut
of dissociated drug was computed it was found not to
differ much for the four compounds. Thus, although the
actual amounts of sulfa drugs which were needed to re-
tard growth were high as compared to those of p-ammo
benzoic acid, the concentrations of dissociated drug were
not very different from those of the acid, their ratio be-
ing in the neighborhood of 1.

Ratio of

Ratio of PABA^ to dissoci-

PABA^ to drug ated drug

Sulfanilamide 1 :5,000 1 :1.4

Sulfapyridine 1: 40 1:1.4

Sulfathiazole 1: 8 1:4.9

Sulfadiazine 1: 8 1:6.4

1. p-amino benzoic acid.

Bell and Roblin (1942) reasoned that the more a sul-
fonamide resembles ^-amino benzoic acid, the greater its
bacteriostatic effect should be. In an extensive study,
they measured the two dissociation constants of 50 sul-
fonamides, and a very definite correlation between acidic
dissociation and antiseptic efficiency was found. ''The
more negative the SO2 group of an N^-substituted sul-
fanilamide derivative, the greater is its bacteriostatic
power."

Kumler and Daniels (1943) point to some inconsis-
tencies in the data of Bell and Roblin, and believe that
the main factor in bacteriostasis is the balance of elec-
tric charges in the molecule. They explain the activity
of the sulfonamides primarily by "the contribution of
resonating form with a coplanar amino group. The neg-
ative character of the SO2 group is a concomitant factor
associated with this resonating form"; it is not the pri-
mary factor as Bell and Roblin thought.

The understanding of the action of sulfonamides is
complicated by the discovery of other antagonizing sub-

156 AXTISEPTICS

stances. Methionine counteracts low concentrations of
sulfonamides, but not larger doses (Kolm and Harris,
1941, 1942). These authors designate as "secondary
antagonists" some other substances with similar effects,
such as xanthine, serine, glycine, which function only
in the presence of methionine.

Several other amino acids have been claimed to be an-
tagonistic to sulfonamide retardation of growth, and
even mercuric chloride can produce this effect, accord-
ing to Lamanna and Shapiro (1943).

Snell and Mitchell (1943) found jDurines to be antagon-
istic to sulfanilamide, but only in the presence of very
small amounts of ^j-amino benzoic acid. Quite generally,
substances which are excellent foods and increase the
growth rate materially, have a slight antagonistic effect
which, however, cannot be considered specific.

The picture of the mode of action of sulfonamides is
still rather hazy and indefinite. But one of their char-
acteristic features seems well established, it is that, in
contrast to other antiseptics, they act more strongly on
growing than on resting cells. The status of the cells is
of great importance, and failure to realize this is prob-
ably the reason for some seemingly contradictory evi-
dence. The kind of medium used is also of importance
since it may contain appreciable amounts of one or sev-
eral antagonists.

Concerning the influence of temperature. White (1939)
found that the efficiency of sulfanilamide is 10 times as
high at 37° as at 30°, and that above 37° it increases
about 10 times for a rise of one degree. This latter in-
crease may be partly due to the harmful effect of high
temperature alone. The minimal bactericidal concentra-
tions were determined with Streptococcus pyogenes incu-
bated in broth, the criteria being cloudiness and presence
of viable bacteria after 48 hours.

AXTISEPTICS 157

VII. INCREASED EFFICIENCY OF WEAK ACIDS
BY INCREASED ACIDITY

It has been known for a long time that salicylic and
benzoic acids are good antiseptics only in acid solution
(see e.g., Cruess and Richert, 1929; Cruess and Irish,
1932). Sulfur dioxide, the standard disinfectant of the
yeast industries, has the same property. Recently, pro-
pionic acid has been added to this group. Boric acid
might also belong in this class, though no records could
be found in the literature to bear out this possibility.
The effect of pH on antiseptic efficiency has been
ascribed by Bittenbender and associates (1940) to a
^ ' specific hydrogen ion effect ' ', while Hoffman, Schweitz-
er and Dalby (1941) explain it by the "relation of polar
and non-polar groups" of the disinfecting acid.

A much simpler explanation has been offered by Rahn
and Conn (1944) who assumed that only the undissoci-
ated acid acts as antiseptic. In neutral solutions, the
acids and their salts are completely dissociated while an
increase of acidity produced by the addition of buffers
of low pH, such as citric acid and acid phosphates, will
decrease the dissociation of the acid in proportion to the
dissociation constant. For benzoic acid, this relation is
given by the equation

[H] [Benzoate ions] ^ ^ ^ ^ ^^^ ^ [H] [a]

[undissociated acid] ' [1 - Q^]

where a represents the "degree of dissociation". The
fraction (1-a), multiplied by the concentration of total
benzoic acid gives the concentration of undissociated acid
at any pH.

The highest concentration still permitting slow growth,
and the lowest concentration preventing growth have
been determined by Rahn and Conn (1944) for yeast.

158

ANTISEPTICS

and are given in Table 32. The preventive dose va-
ried from 50 mg. of sodium benzoate per 100 cc. of
medium, at pH 3.5, to over 2,500 mg., at pH 6.5. These
values, multiplied by 0.847\ give the amounts of free
benzoic acid, and the figures thus obtained multiplied by

TABLE 32

Inhibition of growth of Saccharomyces ellipsoideus by benzoic and
salicylic acid. (From Rahn and Conn, 1944. Courtesy of "Industrial
and Engineering Chemistry.")

pH

Undissociated
fraction
1 —

Inhibitory concentration, in mg. per 100 cc.

Sodium salt
Growth No growth

Undissociated acid
Growth No growth

Sodium Benzoate

3.5

0.83

25

50

17

35

3.65

0.77

20

35

13

24

4.1

0.55

40

50

19

23

4.4

0.38

50

100

16

31

5.0

0.13

250

500

28

55

5.7

0.027

250

500

6.7

11.5

5.8

0.022

1,000

1,500

19

28

6.5

0.003

>2,500

>6.4

Sodium Salicylate

2.93

0.527

10

15

4.5

6.8

4.10

0.070

80

100

4.8

6.0

5.14

0.0068

600

800

3.5

4.8

5.70

0.00188

1,000

2,000

1.6

3.2

the undissociated fraction (1-a), which is characteristic
for a given pH, furnish the concentrations of undisso-
ciated acid. The last column of Table 32 shows that,
even with great differences in the amount of added ben-
zoate, the amount of undissociated acid remains nearly

1. Ratio of molecular weight of benzoic acid to that of sodium
benzoate.

ANTISEPTICS 159

the same, namely about 25 mg. per 100 cc. Evidently,
yeast ceases to multiply when the undissociated acid
reaches a certain concentration.

Salicylic acid has a dissociation constant of 1.06X10'^,
and consequently, its efficiency increases rapidly by a
decrease of pH below 4, while with benzoic acid which
is almost completely undissociated at this acidity, the
efficiency is changed but little by a further decrease in
pH. But the general principle as to the effect of undis-
sociated molecules is the same with both acids as may
be seen from Table 32.

The case of sulfurous acid, which is dibasic and has
two dissociation constants, is somewhat different. We
have

For each pH there exist corresponding concentrations of
HSOs ions, of SO3 ions and of undissociated H0SO3 which
are shown in Table 33.

The experiments with Bact. coli showed very consis-
tently that the growth of this organism is inhibited by a
concentration of about 10 mg. of HSO3 ions per 100 cc.
The growth of yeast, however, was not inhibited by these
ions, but only by the undissociated H0SO3, at concentra-
tions of 0.4 mg. per 100 cc, or 4 ppm.

Rahn and Conn measured also the bactericidal effi-
ciency of H2SO3 and found that it is due exclusively to
the undissociated acid. It is remarkable that Bacterium
coli, of which the growth is so readily inhibited by HSO3
ions, can tolerate about 10 times as much undissociated
acid as yeast.

Propionic acid has recently been used as a fungicide
in the food industries. According to Kulman (1940), it
is more efficient than formic or acetic acid with yeast, but
not with bacteria. Ingle (1940) found the free acid more

160

ANTISEPTICS

TABLE 33

Inhibition of growth by sodium sulfite. (From Rahn and Conn,
1944. Courtesy of "Industrial and Engineering Chemistry.")

PH

Fractions of

SO3
ions

HSO,

Undiss.

Inhibitory concentration,
in mg. per 100 cc.

Na.SOo

Undissoci-
ated

Bacterium coli

5.9 -5.92

0.804

0.196

10-5

60-80

7.6-10.0

0

6.12

0.870

0.130

0

>100

> 8.5

0

6.40

0.926

0.074

0

ca.200

ca. 9.6

0

6.80

0.969

0.031

0

>500

>10.0

0

6.75-7.00

0.968
0.970

0.032
0.020

0

500-1000

10.4-12.8

0

7.05-7.25

0.975
0.980

0.019
0.014

0

1000-2000

12.0-18.0

0

7.6

0.996

0.004

0

>2000

> 5.2

0

7.6 -^7.8

0.997

0.0032

0

2000-5000

5.2-10.5

0

8.0

0.998

0.002

0

>5000

> 6.5

0

Saccharomyces ellipsoideus

1.0

0

0.105

0.855

2.0

0

0.630

0.370

2.48

0.001

0.836

0.163

<5

< 2.7

0.53

2.94

0.005

0.932

0.063

5-10

3-6

0.21-0.41

4.0 -4.1

0.047

0.948

0.0056
0.0050

100-120

68-74

0.40-0.40

4.97

0.332

0.667

0.004
10-5

1200-1400

518-605

0.32-0.37

6.08

0.858

0.142

10

ca.1600

ca.l48

ca.0.01

efficient tlian the sodium or calcium salts. Olson and
Macy (1940) point out that ''the final pH is an extreme-
ly important factor in restraining mold growth". While
2% sodium propionate had no effect at pH 7.0, it suf-
ficed to prevent mold growth for 5 days at pH 6.1; These
authors advance the theory that inhibition may be due
to the undissociated acid.

A\TUSEPTI(\'< 161

This interpretation seems probable as the dissociation
constant of propionic acid is 1.3 X 10\ The fractions
of undissociated acid are

at pH 5 0.43

at pH 6 0.072

at pH 7 0.0076

at pH 8 0.0007.

In neutral or alkaline media, the fraction of undissoci-
ated acid is so small that even a high concentration of
proprionate would give only a very small amount of acid.
But when the s.olution becomes acid, the undissociated
portion increases rapidly from less than 1% at neutral-
ity to 43% of the total propionate at pH 5.

A similar effect by the undissociated fraction of an
acid has been observed by Rogers and Whittier as early
as 1928. They found that the lactic fermentation by
Streptococcus lactis ceases, regardless of pH, when the
concentration of undissociated lactic acid becomes more
than 0.017 molar.

The strong action of the undissociated acids as com-
pared with that of their ions may be explained by the
ability of the undissociated molecules to pass through
the membrane and through protoplasm "with great rap-
idity, according to the simple laws of diffusion" while
''the permeability of cells to ions is a decidedly complex
phenomenon, more or less limited in its extent, and in-
volving theoretical equilibria which may be approached
very slowly, and perhaps never be attained" (Jacobs,
1940).

Benzoic acid and sulfur dioxide cause an increased lag
period as well as a decreased multiplication rate. Figure
31 shows the growth of yeast at pH 4.25, with increasing
amounts of sodium benzoate. The curves with NaoSOs
are similar.

The length of the lag period and the rate of multipli-
cation depend on pH as well as on the concentration of the

162

ANTISEPTICS

Fig. 31. Multiplication curves of ^acvharomj/crs ellipsoideus treated
with various concentrations of benzoic acid, at pH 4.1.

antiseptic. This is shown graphically in Fig 32 which
represents the effect of SO2 on Bacterium coli. The con-
centration of the added NaoSOs is plotted logarithmically
against the pH. The slanting dotted line is the border
line of tolerance. Above this line, the lag period is
infinite, and so is the generation time. Just below this
line, bacteria require 30 to 140 hours before growth be-
comes visible, and 2 to 4 times as many hours for doub-
ling their number. The generation times in this figure are
only relative values because multiplication was measured
by turbidity.

A similar set of data for yeast with benzoic acid (Fig-
ure 33) where multiplication had been determined by
direct counts and plate counts, represents true generation
times. Here also, the lag period as well as the multipli-
cation rate are greatly affected. The line representing
the limit of tolerance has a slope different from that in
Figure 32.

AXThSEr'TIC\S

163

•, 35:32

-:is —^ —

5.5 6 p^

Lag period in hours

0.06%
0.04%

1

40-

—

-,

— e. —

■ a —

9

1 • — 1

-

IS

-

—

"^l^t**

, i^

3

-7.5

-

— 7 ■

-

6

7 Ph 8

Relative generation time

Fig. 32. Effect of sodium sulfite on the lag period (left), and on
the generation time (tight) of Bacterium coli at different acidities.
The broken line represents the borderline of tolerance, above which
the lag period or the multiplication time are infinite and below which
they have the values indicated.

> PH

1.9% Ui
1.0% <

0.3% m
025%!

0 03%
002%

PH ,6

Lag period in hours

Generation time

Fig. 33. Effect of sodium benzoate on the lag period (left) and on
the generation time (right) of yeast at different acidities. The broken
line represents the borderline of tolerance, above which the lag period
or the generation time are infinite and below which they have the
values indicated.

1 64 ANTISEPTICS

VIII. ANTAGONISTIC SUBSTANCES FROM MICROORGANISMS

Just as higher ammals are protected against invasion
of microorganisms by the secretion of antibodies, lyso-
zymes, and the bactericidal compounds of blood, saliva
and milk, so some microorganisms prevent the intrusion
of others into their ''Lebensraum" by secreting toxic
substances. Certain notable antagonisms between diff-
erent species have been known for a long time. As early
as 1887, Garre reported the inhibition of staphylococci
and typhoid bacteria by certain fluorescent species. The
pyocyanase of Pseudomonas aeruginosa {pyocyanea)
has been studied very extensively. In recent years, much
attention has been paid to antibiotic substances from mi-
croorganisms in the hope that they might be useful in
chemotherapy (see review by Waksman, 1941). This hope,
however, had to be abandoned in the case of most of
these substances because of their toxicity to man. Only
gramicidin and penicillin have been retained on the list
of the probably useful bacterial products but the search
for similar compounds continues.

The long list of organisms which produce substances
toxic to other microorganisms includes hyphomycetes,
actinomycetes, sporulating and non-sporulating bacteria,
but no yeasts and no protozoa. While most groups of
microorganisms are represented in this list, there are
only a few species in each group.

Concerning the chemical nature of antibiotic agents,
much remains to be investigated. There is enough evi-
dence, however, to state that they belong to quite dif-
ferent chemical groups. Dubos (1935) described an
enzyme, which dissolves the capsule of pneumococci and
is so specific that it attacks only Type III. On the other
hand, the substance called pyocyanase, which was as-
sumed to be an enzyme, is probably a lipoid. Gramicidin

ANTISEPTICS

165

and tyrocidin, two different substances produced by the
same Bacillus hrevis, have been obtained in crystalline
form by Dubos (1939-40) and have molecular weights
of about 900 to 1400 but their structure has not yet been
determined. The penicillin of Penicillhim notatum is so
unstable that its structure is also unknown. This sub-
stance retains its activity only at low temperatures ; how-
ever, stable and fairly efficient esters have been obtained
(Meyer et al., 1943)/

It is not probable that substances so different in na-
ture will act upon bacteria in the same manner. How-
ever, little can be said about their mode of action. They
cannot be included in any of the groups of antiseptics
already discussed because their chemical nature is too
uncertain. Most of them are highly selective, as has been
stated above for gramicidin and penicillin (they are
good antiseptics, but not strong disinfectants). Penicil-
lin will inhibit a few species at a concentration of 1 ppm.,
but others not even at 1000 ppm. (Abraham et al., 1941 )^
Table 84 gives some of the data by Waksman and Wood-

TABLE 34

Growth-inhibiting concenti-ations (in parts per million) of various
anti-biotic substances. (From Waksman and Woodruff, 1942.)

Bad. coli

B. mycoides

B. subtilis

Sarcina lutea

Actinomycin

300

0.1

0.03

0.3

Streptothricin

3

> 300

1

30

Tyrothricin*

>3000

30

30

10

Gramicidin

> 300

> 300

> 300

10

Tyrocidin

> 300

3

10

3

Pyocyanase

> 300

30

30

30

Pyocyanin

30

10

3

10

Penicillin

> 300

> 300

1

1

Gliotoxin

100

10

3

0.1

Lysozyme

> 100

100

3

0.1

Phenol

3000

3000

1000

1000

Sulfanilamide

1000

>1000

100

100

*Total active material from Bacillus hrevis, containing gramicidin
and tyrocidin. ,..-—

1. See also p. 124.

C^fi

<ifr4>*'!&' V>

(4J

%■

LlS?<LAftY

S9

vf

166 ANTISEPTICS

ruff (1942) concerning the concentrations required for
inhibition of growth of four different organisms.

Staphylococci and streptococci can be adapted to very
high doses of penicillin, 30 times the originally inhibiting
concentration for the streptococci, 1000 to 6000 times for
the staphylococci (Abraham et al., 1941; McKee and
Houck, 1943). In this process of adaptation, increase in
tolerance was combined with an almost complete loss
of virulence. These projDerties were retained after trans-
fer to normal broth, which is contrary to the usual prompt
loss of acquired tolerance. The same retention of peni-
cillin-resistance has been reported by Schmidt and Sesler
(1943) for Pneumococcus Type III. There is no relation
between resistance to penicillin and to sulfonamides.

The disinfectant action of penicillin has been studied
by Hobby, Meyer and Chaffee (1942). The order of
death of pneumococci, streptococci and staphylococci by
50 ppm. of penicillin was logarithmic, at least during the
destruction of the first 99% of the cells (Figure 34). The
death rate decreased when the size of the inoculum in-
creased, w^hicli may be due to the protective action of large
numbers of cells. The penicillin concentration was not
noticeably decreased as a result of its bactericidal action.
Death was not due to Ij^sis. In a full grown, 18-hour-
old culture of Streptococcus hemolyticus, 100 ppm. peni-
cillin did not decrease the number of viable cells; but
with small inocula, after the bacteria w^ere given a chance
to multiply, ths same concentration killed almost all the
cells at 37°, a smaller number died at 18°, and none at
4°. The deathrate constants were 0.100 and 0.027, re-
sulting in the very low temperature coefficient Qio = 1-9.

Remarkable morphological changes are brought about
by penicillin, which give a clue in regard to the mode
of attack of this drug. Gardner (1940) observed that
Clostridium welchii, which is completely inhibited by 17
ppm. penicillin, showed, when treated w^ith as little as 1
ppm. ''an extreme elongation of the majority of the cells.

ANTISEPTICS

167

Fig. 34. Left: Death curve of various cocci treated with oO ppm. of
penicillin. Abscissa: time in hours; ordinate: logarithm of the number
of cells per cc.

Right: Relation between penicillin concentration and death time of
Streptococcus hemolyticus. Abscissa: death time in hours, on logarith-
mic scale; ordinate: penicillin concentration, in ppm., on logarithmic
scale. (From data of Hobby, Meyer and Chaffee, 1942.)

which took the form of unsegmented filaments ten or
more times longer than the average cell." Gram-nega-
tive rods also became greatly elongated. Vibrio comma
grew into immensely swollen filaments. ''With staphylo-
cocci, the morphological change takes the form of spher-
ical enlargement of the cell and imperfect fission . . .
Streptococcus pyogenes showed great swelling of the
cells, incomplete fission with formation of large spinoles,
and increased length of chain." Gardner concludes:
"Growth proceeds, but division and separation do not
follow in due course. Many cells then fall victim to
autolysis." In the experiments mentioned in Table 35,
the present writer observed in the streptococcus cultures
giant cells and rod-shaped cells, apparently due to con-
tinued growth in one dimension without formation of par-
titions, but litle change was noted in the staphylococcus
cultures.

Similar morphological changes have been occasionally
reported for other antiseptics, e.g., for methyl violet by
Walker and Murray (1904) and for sulfanilamide by
Tunnicliff (1939). But if such effects were the rule
with those substances they would have been more fre-
quently reported.

168

AXTISEPTICS

Gardner's observations remind one of the effect of col-
chicin upon plant cells.

These morphological effects are in line with the ob-
servation by Eagle (1944) that "penicillin impairs the
viability of cultured spirochetes, as determined by their
failure to grow in subculture, long before their motility
has been demonstrably affected. There is at least a hun-
dred-fold disparity between the concentrations of peni-
cillin necessary to immobilize organisms in a given time
period, and those which render the organisms non-
viable. ' '

Waksman and Woodruff (1942) suggest that our cus-
tomary media may contain a substance inhibitory to anti-
biotic action. That would indicate a mode of action
similar to that of the sulfonamides. However, the dif-
ferences of inhibition in synthetic and in other complex
nutrient media are so small that they do not offer con-
vincing proof for this assumption.

Welshimer, Krampitz and Werkman (1944) observed
an interference of penicillin with the dismutation of py-

TABLE 35

Effect of penicillin on the multiplication and final crop of various
eria. (Original data.)

bactei

Final crop

(Relative turbidity

Time (in

hours) required to

values)

increase the turbidity
10 per cent

Penicillin in
glucose broth

Large

Small

(ppm.)

inoculum

inoculum

Strept.
lactis

Staph,
aureus

Bad.
coli

Streptococcus

0

0.7

0.8

0.9

86

85

1.25

0.8

14.0

1.0

72

78

2.5

1.3

00

1.2

24

29

5

8.5

0.95

(76)

(81)

10

>100

1.1

5

7

15.6

X

1.2

0

0

62.5

1.0

0

0

250

1.0

0

0

ANTISEPTICS 169

ruvate, but no concentrations are mentioned in their very
brief abstract.

The experiences of Gardner and of Eagle mentioned
above leave little doubt that the mechanism of cell divi-
sion is affected by concentrations which interfere neither
with growth nor with motility.

Multiplication curves of bacteria in presence of peni-
cillin were studied by the writer who measured neophelo-
metrically the retardation of multiplication with several
types of bacteria. No definite increase in the lag period
could be observed, and all retardation seemed to be due
to a decreased growth rate (Table 35). Streptococcus
lactis in presence of 10 ppm. penicillin never multiplied
enough to increase the turbidity 10%. The final crops
show the usual decrease with increasing antiseptic con-
centration (except wdth 5 ppm. where multiplication for
the first 3 days was in conformity with that of the other
cultures, while its velocity increased on the fourth day;
the increase could not be traced to a contamination).

From these growth studies, the action of penicillin
seems to be identical with that of many antiseptics.
Penicillin does not act like a dye nor like a sulfonamide.

The understanding of the nature of the penicillin effect
is complicated by the fact that at least two different bac-
tericidal substances are produced by the mold. Besides the
true penicillin, a penicillin 5 has been isolated by Rob-
erts et al. (1943), while Kocholaty (1942) discovered a
substance called penatin, and Coulthard et al. (1942)
found an enzyme which they called notatin. This en-
zyme is an aerobic dehydrogenase oxidizing glucose to
gluconic acid with the simultaneous production of HoOo.
Birkinshaw and Raistrick (1943) assume that penicillin
B, penatin, and notatin are the same substance, which
acts simply by gradually accumulating an inhibiting
amount of HoO^. This substance may completely inhibit
multiplication in a dilution of 1 part in a billion, but only
when glucose and oxygen are present, and catalase absent.

SUMMARY AND CONCLUSIONS

1. The first and immediate effect of toxic substances
upon bacteria is an interruption of their multiplication.
Reproduction may be only retarded or it may be com-
pletely inhibited, dejDending upon the concentration of
the toxic substance. If the concentration is sufficiently
low, the cells may not be damaged, the action of the toxic
substance on them is reversible, multiplication becoming
normal after this substance is removed. Such a reversi-
ble inhibition of multiplication, without death, character-
izes antisepsis. The condition thus defined is, how^ever, an
ideal one, rarely realized; usually a certain amount of
slow death accompanies antisepsis. With higher con-
centration of the toxic substance, another effect is added
which characterizes disinfection. The cells lose their re-
productive function permanently because of an irrever-
sible reaction of the toxic substance with some cell con-
stituent. Permanent loss of reproductive power is to
the bacteriologist equivalent to the death of the cell.

2. The antiseptic action is instantaneous, except with
the sulfonamides which permit a fairly normal multi-
plication for 2 or 3 generations before the toxic effect is
noticeable. The disinfectant action is not instantaneous,
but proceeds at a measurable rate, called the death rate
(the percentage of cells dying during each unit of time).

3. As a rule, the death rate is constant throughout an
experiment, and this results in a ''logarithmic order of
death." In applications to chemical disinfection, the
''deathrate constant" is a valuable tool for the study
of the fundamental lethal reaction.

4. A constant death rate indicates that death is caused
by the reaction of one single molecule of the cell with the

CONCLUSIONS 171

disinfectant. It seems logical to assume that this all-
important molecule is either a gene or a similar master-
molecule of the cell division mechanism. Several authors
have defined the death of bacteria as a lethal mutation.
In the outstanding excej^tional case of chlorine, for
which the order of death is not logarithmic, it is prob-
able from microscopic observation that the cell membrane
is attacked first, and that many membrane molecules
must be destroyed before the cell is dead.

5. Constant death rates permit us to express the ef-
fects of temperature and of concentration in simple
terms. The death rate increases with increasing tem-
perature, and, as a rule, the temperature coefficient for
10 degrees C. is between 3 and 5.

6. The efficiency of disinfectants increases with their
concentration, but not in the same manner for all disin-
fectants. With formaldehyde, hydrogen peroxide and
many other disinfectants, the deathrate constant is ap-
proximately proportional to the concentration, but with
some others, it is proportional to the square of the con-
centration, or to the square root, and with the phenols it
increases as the 6th to 8th power of the concentration.

7. The only standard method for the evaluation of
disinfectants is the phenol coefficient which states how
many times as powerful as phenol a disinfectant is. As
the efficiency of phenol changes with the concentration
In an exceptional manner, different from that of prac-
tically all other antiseptics, this ratio varies greatly when
dilute concentrations are compared. The choice of phenol
as a standard is therefore most unfortunate. The range
allowed by the Federal Drug Administration Method
gives values with a calculated permissible deviation of
260%. A new method is projDosed here which gives the
death times for any concentration of the disinfectant di-
rectly, not merely in comparison with phenol. By this
method, the interference of organic matter with disin-
fection can also be measured quantitatively.

172 CONCLUSIONS

8. The efficiency of antiseptics cannot be studied by
the technique used with disinfectants. As the reaction
of antiseptics is instantaneous, no rate of reaction can
be computed, and the mode of action can be ascertained
only by observing the effects of weaker doses which re-
tard, but do not stop multiplication. No temperature
coefficient and no concentration exponent can be estab-
lished for antiseptics.

9. Antiseptics retard multiplication in two quite dif-
ferent ways; they may decrease the multiplication rate,
or they may increase the lag period, i.e., the time normal-
ly required by transferred cells to adjust themselves to
the new environment, or to change from the senile to the
juvenile state. Dilute phenol, penicillin, and sulfon-
amides decrease the reproduction rate without affecting
the lag period. Dyes extend the lag period, eventually
to several days, without reducing the rate of reproduction,
if multiplication starts at all. Formaldehyde, sulfur diox-
ide and benzoic acid prolong the lag phase and also re-
duce the rate of multiplication.

10. As was said above, the lethal reaction of most dis-
infectants is a reaction with some molecule without which
cell division cannot take place. It seems certain that this
molecule must be of protein nature. No other single
molecule can be so important. The compounds causing
death belong to a great variety of chemical groups, and
do not react in the same way with proteins. But a native
protein has many very different sidechains, and the inac-
tivation of any one of these may be sufficient to terminate
its physiological function. Thus the fundamental lethal
reaction of different disinfectants may not be identically
the same, but it represents an attack on the same mole-
cule, or group of molecules. Exceptions must be ex-
pected. Chlorine has already been mentioned as attack-
ing first the cell membrane. It is imaginable that some
compounds destroj^ the enzymes more readily than the
synthesis or multiplication mechanisms, though all evi-

CONCLUSIONS 173

dence, so far, has shown that enzymes react more slowly
than the reproduction mechanisms.

11. Growth retardation by antiseptics is not always due
to the same reaction. Here too, as a rule, the enzymes
suffer less than the mechanisms of synthesis and cell di-
vision. But there is more variety in the mode of attack
than with disinfectants. The most common effect is a
decrease of the growth rate, probably from partial inacti-
vation of the catalysts which bring about synthesis. The
sulfonamides are different from all other antiseptics, for
they do not affect resting cells. They interfere only with
the third or fourth generation of growing cells, and the
interference can be avoided completely by addition of
p-amino benzoic acid. This suggests that growth is not
inhibited because the sulfonamide reacts with some essen-
tial cell protein, but because it becomes part of the cell
structure which then fails to function.

12. The retardation by dyes is quite certainly due to
an abnormal oxidation-reduction potential, and if the
cells are present in sufficient numbers, they can readjust
the potential and can then multiply completely unre-
stricted.

13. An investigation of the relation between the acidity
and the antiseptic power of organic acids and sulfur diox-
ide has shown that the undissociated acid molecules are
the. real inhibitory factor, probably because they get into
the cell more readily than the ions.

14. The picture of death and of growth inhibition is
by no means perfectly clear, and it cannot be clear as
long as we have no perfect understanding of the mechan-
isms upon which life depends. But by varying the con-
dition of death, by varying the species under test, and by
applying quantitative methods, our conceptions of the
death of the cell are becoming more definite, and ulti-
mately this study will not only explain the mechanism of
death, but will permit important conclusions about the
unknown mechanism of life.

BIBLIOGRAPHY

ABRAHAM, E. P., E. CHAIN, C. M. FLETCHER, A. D. GARDNER,

N. G. HEATLY, M. A. JENNINGS and H. W. FLOREY, Lancet,

2^1, 177, 1941.
ANDERSON, L. P. and W. L. MALLMANN, Mich. State College Tech.

Bull, 183, 1943.
BAKER, Z., R. W. HARRISON and B. F. MILLER, J. Exp. Med., 73,

249, 1941; 74, 611 and 621, 1942.

BARRON, E. S. G. and H. R. JACOBS, Proc. Soc. Exp. Biol, and Med.,

37, 10, 1937.
BEAMER, P. R. and F. W. TANNER, Zbl. f. Bakt., II, 100, 81, 1939a;

100, 202, 1939b.
BECHHOLD, H., Ztschr. f. Electrochem., 2.'t, 147, 1918.
BELEHRADEK, J., Temperature and Living Matter, Berlin, 1935.
BELL, P. H. and R. O. ROBLIN, J. Am. Chem. Soc, 6^, 2905, 1942.
BIRKHAUG,.K. E., J. Infect. Dis., 53, 250, 1933.
BIRKINSHAW, J. H. and H. RAISTRICK, J. Biol. Chem., UfS, 459,

1943.
BITTENBENDER, W. A., E. F. DEGERING, P. A. TETRAULT, C. F.

FEASLY and B. H. GWYNN, Ind. and Eng. Chem., 32, 996, 1940.

BORMAN, E. K., J. Bad., 23, 315, 1932.
BOYCOTT, A. E., J. Path, and Baet.. .?.;. 237, 1920.
BRONFENBRENNER, J., A. D. HERSHEY and J. A. DOUBLY, Proc.

Soc. Exp. Biol, and Med., 38, 210, 1938.
CAMERON, E. J., J. Bact., 19, 52, 1930.
CHARLTON, D. and M. LEVINE, loiva State College Engineering Exp.

Sta. Bull, 132, 1937.
CHICK, H., J. Hyg., 8, 92, 1908; 10, 237, 1910.

and C. J. MARTIN, J. Hyg., 8, 654, 1908.

, J. Physiol, 40, 404, 1910.

CHU, H. I. and A. B. HASTINGS, J. Pharmacol, 63, 407, 1938.

CHURCHMAN, J. W., J. Exp. Med., 16, 221, 1912.

CLARK, H. W. and S. de M. GAGE, 34th Ann. Report, State Board of

Health Massachusetts, p. 245, 1903.
COHEN, B., J. Bact., 7, 183, 1922.
COOPER, E. A. and J. MASON, J. Hyg., 26, 118, 1927.

, J. Path, and Bact., 31, 343, 1928.

COULTHARD. E. E., R. MICHAELIS, W. F. SHORT, G. SYKES, G.

E. H. SKRIMSHIRE, A. F. B. STANDFAST, J. H. BIRKINSHAW

and H. RAISTRICK, Nature, 1.50, 634, 1942.
CRUESS, W. V. and J. H. IRISH, J. Bact., 23, 163, 1932.

and P. H. RICHERT, J. Bad., 17, 363, 1929.

CURRAN, H. R., F. R. EVANS and A. LEVITON, J. Bad., 40, 423,

1940.
DAVEY, W. P., J. Exp. Zool, 22, 573, 1917.
DIERNHOFER, K., Ztschr. Fleisch und Milch Hygiene, 43, 328, 1933.

, Milchw. Forschungen, IS, 83, 1936.

DITTMAR, H. R., I. L. BALDWIN and S. B. MILLER, J. Bact., 19,

203, 1930.

BIBLIOGRAPHY 175

DORFMAN. A.. L. RICE. S. A. KOSER and F. SAUNDERS. Proc.

Soc. Exp. Biol, and Med., ^5, 750, 1940.

and S. A. KOSER, J. Infect. Dis., 11, 241, 1942.

DOWNS, C. M., J. Bad., U, 392, 1942.

DUBOS, R., J. Exp. Med., 49, 575, 1929; 62, 259, 1935; 70, 1, 1939.

and R. D. HOTCHKISS, J. Exp. Med., 13, 629, 1941.

EAGLE, H., J. Bad., 47, 427, 1944.

EGGERTH, A. H., J. Exp. Med., 46, 671, 1927; Jf9, 53, 1929; 50, 299, 1930;

53, 27, 1931.
EIJKMAN, C, Versl. Kon. Akad. Wete^ischappen Amsterdam, Wis-en

Natuu'rkunde, 21 (I), 507, 1912.
ELY, J. 0., J. Bad., 38, 391, 1939.

ESTY, J. R. and K. F. MEYER, J. Infed. Dis., 31, 650, 1922.
FILDES, P., British J. Exp. Path., 21, 67, 1940.
FOTER, M. J. and O. RAHN, J. Bad., 32, 485, 1936.
FOX, C. J. and H. M. ROSE, Proc. Soc. Exp. Biol, and Med., 50,

142, 1942.
FRICKE, H. and M. DEMEREC, Proc. Nat. Acad. Science, 320, 1937.
GAGE, S.de M. and G. STOUGHTON, Science, 23, 216, 1906.
GARDNER, A. D., Nature, 146, 837, 1940.
GARRe, C, Zbl f. Bakt., 2-, 312, 1887.
GATES, F. L., J. Gen. Physiol., 13, 231, 1929.
GEGENBAUER, v., Archiv f. Hygiene, 90, 23, 1921.
GREIG, M. E. and J. C. HOOGERHEIDE, J. Bad., 41, 557, 1941.
GROVES, J. F., Bot. Gazette, 63, 169, 1917.
HARTMANN, G., Milchw. Forschungen, 18, 116, 1936.
HARVEY, H. W., Ann. Bot., 23, 181, 1909.
HASTINGS, E. G., E. B. FRED and W. R. CARROLL, Zbl. f. Bakt., II,

67, 162, 1926.
HENRY, R. J., Bact. Revieivs, 1, 176, 1943.
HEVn:.ETT, R. T., Lancet, 741, 1909 (I).
HIRSCH, J., Com.pt. rend. Soc. Turque, Sciences Physiques et Natur-

elles, 10, 19, 1942
HOBBY, G. L., K. MEYER and E. CHAFFEE, Proc. Soc. Exp. Biol.

and Med., 50, 281, 1942.
HOFFMAN, C, T. R. SCHWEITZER and G. DALBY, Ind. and Eng.

Chem., 33, 749, 1941.
HOFFMANN, C. E., Thesis, Cornell University, 1943.

and O. RAHN, J. Bact., 41, 177, 1944.

HOLWERDA, B. J., Mededeelingen van den Dienst der Volksgeuzond-

heid in Ned. Indie, 11, 251, 1928; 19, 326, 1930. (Quoted from Charl-
ton and Levine, 1937)

INGLE, J. D., J. Dairy Science, 23, 509, 1940.

INGRAHAM, M. A., J. Bad., 26, 573, 1933.

ISAACS, M. L., In F. P. Gay, Agents of Disease and Host Resistance,
p. 228, 1935.

JACOBS, M. H., Cold Spring Harbor Symposia, 8, 30, 1940.

JOHNSON, F. H., Science, 95, 104, 1942.

and K. MOORE, Proc. Soc. Exp. Biol, and Med. 48,

323, 1941.

176 BIBLIOGRAPHY

JOHNSON, J. R., W. F. BRUCE and M. A. DUTCHER, J. Am. Chem.

Soc, 65, 2005, 1943.
KATZIN, L. I. and L. A. SANDHOLZER, J. Bad., J,Jt, 722, 1942.
KLARMANN, E., Soap, 9, Dec. 1933.

and V. A. SHTERNOV, Soap and San. Chem., 17, Jan.

1941.

KNAYSI, G., /. Infect. Dis., .',7, 293, 303, 318, 322, and 328, 1930.

KOCHOLATY, W., J. Bad., //'/, 469, 1942.

KOHN, H. I. and J. S. HARRIS, J. Pharmacol., 73, 343, 1941; 73, 383,

1941a.

, J. Bact., U, 717, 1942.

KRAUSE, G. A. and J. F. BERGMANN, Neiie Wege zur Wasserster-

ilisierung (I-'atadyn). Miinchen, 1928.

KRoNIG, B. and TH. PAUL, Ztschr. f. Hygiene, 25, 1, 1897.

KULMAN, J., Chem. Listy, 33, 420, 1939. (Quoted from Chem. Ab-
stracts, No. 5471, 1940.)

KUMLER, W. D. and T. C. DANIELS, J. Am. Chem. Soc, 65, 2190,
1943.

LAMANNA, C. and I. M. SHAPIRO, J. Bact., ^5, 385, 1943.

LANDY, M., N. W. LARKUM, E. J. OSWALD and F. STREIGHTOFF,
Science, 97, 265, 1943.

LANGE, B., Ztschr. f. Hygiene, 96, 92, 1922.

LEE, R. E. and C. A. GILBERT, J. Phys. Chem., 22, 348, 1918.

LEE, S. W., J. A. EPSTEIN and E. J. FOLEY, Proc. Soc. Exp. Biol,
and Med., 54, 107, 1943.

LEONARD, G. F., J. Infect. Dis., 48, 359, 1931.

LLOYD, L., Ann. Applied Biol., 7, 66, 1920.

LOEB, J. and J. H. NORTHROP, J. Biol. Chem., 32, 103, 1917.

LOMMEL, J., Comjyt. Rend. Soc. Biol., 95, 714, 1926.

MacLEOD, C. M., Proc. Soc. Exp. Biol, and Med., 41, 215, 1939.

and G. S. MIRICK, J. Bact., 4^, 277, 1942.

MADSEN, T. and M. NYMAN, Ztschr. f. Hygiene, 57, 388, 1907.
MAGOON, C. A., J. Bact. 11, 253, 1926.

MALLMANN, W. L. and W. B. ARDREY, Mich. State College, Eng.

Exp. Sta. Bull., 91, 1940.
McCULLOCH, E. C, Disinfection and Sterilization. Philadelphia, 1936.
McILWAIN, H., Brit. J. Exp. Path., 21, 136, 1940.

, Lancet, 412, 1942 (I).

McINTOSH, J. and L. E. H. WHITBY, Lancet, 431, 1939 (I).

McKEE, C. M. and C. L. HOUCK, Proc. Soc. Exp. Biol, and Med., 53,

33, 1943.
HEADER, P. D. and W. A. FEIRER, J. Infect. Dis., 39, 237, 1926.
MELLON, R. R. and L. L. BAMBAS, Proc. Soc. Exp. Biol, and Med.,

36, 682, 1937.
MEYER, K., G. L. HOBBY and E. CHAFFEE, ,Science, 97, 205, 1943.
and M. H. DAWSON, Proc. Soc. Exp. Biol, and

Med., 53, 100, 1943.
MILLER, B. F., R. ABRAMS, D. A. HUBER and M. KLEIN, Proc.

Soc. Exp. Biol, and Med., 54, 174, 1943.

BIIilJOGRAFHY 1/7

MUIR, R. D., V. J. SHAMLEFFER and L. R. JONES, J. Bad., U,

95, 1942.
MuLLER, A., Archiv f. Hygiene, 89, 363, 1920.
MYERS, R. P., /. Bact., 15, 391, 1928.

, J. Agr. Research, 38, 521, 1929.

O'KANE, W. C, W. A. WESTGATE and L. C. GLOVER, Netv Hampshire

Agr. Exp. Station Tech. Bull., 5S, 1934.
OLSON, J. C. and H. MACY, J. Dairy Science, 23, 509, 1940.
PAUL, T., Biochem. Ztschr., 18, 1, 1909.
PAUL, TH., G. BIRSTEIN and A. REUSS, Biochem. Ztschr., '2<J, 202

and 249, 1910.
PENFOLD, W. J., Proc. Roy. Soc. Med., 'to, 97, 1911.
PETERS, R. A., J. Physiol., 54, 260, 1920.
PHELPS, E. B., J. Infect. Dis., 8, 27, 1911.
RAHN, O., J. Gen. Physiol., 13, 179, 1929; 13, 395, 1930; 14, 315, 1931.

, Physiology of Bacteria, Philadelphia, Blakiston, 1932.

, Cold Spring Harbor Symposia on Quantit. Biol., 2, 70, 1934.

and M. N. BARNES, J. Gen. Physiol., 16, 579, 1933.

. and F. M. BIGWOOD, Arch. f. MikroMol., 10, 1, 1939.

and J. E. CONN, Ind. and Eng. Chem., 36, 185, 1944.

RALSTON, A. W. and C. W. HOERR, J. Organic Chem., 7, 546, 1942.

REICHEL, H., Biochem. Ztschr., 22, 149, 1909.

REICHENBACH, H., Ztschr. f. Hygiene, 69, 171, 1911.

REVIS, C, Proc. Roy. Soc, B, 85, 192, 1912.

RIDEAL, S. and J. T. A. WALKER, /. Roy. San. Inst., 24, 421, 1903.

ROBERTS, E. C., C. K. KAIN, R. D MUIR, F J. REITHEL, W. L.
GABY, J. T. VAN BRUGGEN, D. M. HOMAN, P. A. KATZMAN,
L. R. JONES and E. A. DOISY, J. Biol. Chem., W, 47, 1943.

ROGERS, L. A. and E. O. WHITTIER, J. Bact., 16, 211, 1928.

SATTLER, W., Milchwirtsch. Forschungen, 7, 100, 1928.

SCHAFFER, J. M. and F. W. TILLEY, J. Infect. Dis., 31, 359, 1925.

, J. Agr. Research, 41, 737, 1930.

SCHMIDT, L. H. and C. L. SESLER, Proc. Soc. Exp. Biol, and Med.,
52, 353, 1943.

SEVAG, M. G. and M. SHELBURNE, /. Bact., 43, 411, 421 and 447,
1942.

SHERMAN, J. M. and W. R. ALBUS, J. Bact., S, 127, 1923.

SLANETZ, L. W. and L. F. RETTGER, J. Bact., 26, 599, 1933.

SMITH, J. HENDERSON, Ann. Applied Biol., S, 27, 1921; 10, 335, 1923.

SNELL, E. E. and H. K. MITCHELL, Arch, of Biochem., 1, 93, 1943.

STARKEY, R. L. and S. A. WAKSMAN, J. Bact., 45, 509, 1943.

STEARN, A. E. and E. W. STEARN, J. Bad., 9, 491, 1924; 11, 345,
1926.

STEINBERG, R. A., Bot. Gazette, 97, 666. 1936.

THOM, C. and R. A. STEINBERG, Proc. Nat. Acad. Science, 25, 329,
1939.

THOMAS, R. C, Ohio Agr. Exp. Sta. Tech. Bull, 10, 1932.

TILLEY, F. W., J. Bad., 38, 499, 1939; 43, 521, 1942.

178 BIBLIOGRAPHY

TILLEY, F. W., and R. M. CHAPIN, J. Bad., 19, 295, 1930.

and J. M. SCHAFFER, J. Infect. Dis., 37, 359, 1925.

-^ , J. Agr. Research, 41, 737, 1930; ^2, 93, 1931.

TUNNICLIFF, E. A., J. Infect. Dis., 6Jf, 59, 1939.

WAKSMAN, S. A., Bad. Reviews, 5, 231, 1941.

and H. B. WOODRUFF, J. Bad., U, 373, 1942.

WALKER, J. E., J. Infect. Dis., 35, 557, 1924; 37, 181, 1925.
WALKER, J. T. A. and R. S. MURRAY, Brit. Med. J., 2, 16, 1904.
WATKINS, J. H., Thesis. Yale University, 1932.

and C. E. A. WINSLOW, J. Bad., 24, 243, 1932.

WATSON, H. E., J. Hyg., 8, 536, 1908.

WELSHIMER, H. J., L. O. KRAMPITZ and C. H. WERKMAN, J. Bact.,

45, 425, 1944.
WHITE, H. J., J. Bact., 88, 549, 1939.

WINSLOW, C. E. A. and 0. R. BROOKE, J. Bad., 13, 235, 1927.
WINSLOW, C. E. A. and I. S. FALK, J. Bad., 11, 1, 1926.
WOODS, D. D., Brit. J. Exp. Path., 21, 74, 1940.
WYCKOFF, R. W. G., J. Exp. Med., 52, 435, 1930.

and T. M. RIVERS, J. Exp. Med., 51, 930, 1930.

WYSS, 0., F. B. STRANDKOV and F. C. SCHMELKES, Science, 96,

236, 1942.
ZEUG, M., Archiv f. Hygiene, SO. 175, 1920.

SUBJECT INDEX

Acids, disinfection by, 76, 82, 84

Adaptation to poisons, 58

Age of cells, effect on death, 42,
141

Alcohols, concentration exponents
of, 73, 89; disinfection by, 89

Algae, death of, 51

Alkalies, disinfection by, 85, 108

Antagonists to sulfonamides, 153

Antibiotics, 164

Antidotes, 82, 86

Antiseptics, definition of, 62, 63.
119; effect of age of cells on ac-
tion of, 141; effect of concentra-
tion of, 120, 132, 141, 148; effect
of cell concentration on, 133,
143; effect of temperature on,
121; evaluation of, 118, 120;
mode of action of, 130, 138; se-
lectivity of, 123

Argyn, argyrol, 88

Aspergillus, death of spores of, 51

Bacteriostasis, 137
Benzoic acid, 157, 162

Cell concentration, and antisepsis,

133, 143; and death time, 54
Cell mechanisms, 9, 10, 11
Chemotherapy, 83
Clilarnydomonas. death of, 51, 52
Chlorine, efficiency of, 80, 82, 92;

disinfection, abnormal order of

death in, 51, 53
Colloidal metals, 88, 89
Colpidium, death of, 51, 52
Concentration exponents, 66, 73,

102; meaning of, 70

Copper salts, 88, 89

Crop decrease by antiseptics, 121,
129, 169

Crystal violet, 127, 138

Dead cells protecting living cells,

58

Death, a mutation, 41; by heat, 14,
15, 18; by X-rays, 14, 15, 18;
cause of, 28, 31; definition of, 9,
40; non-logarithmic order of,
42; order of, 13, 28, 31; order
of, when very slow, 47

Deathrate constant, 28, 30

Deathtime, 30

Decimal reduction time, 30

Detergents, 99

Disinfectants, as antiseptics, 128,
132; concentration exponents
of, 66; definition of, 62, 63, 119;
efficiency of, 101, 109; interfer-
ence by foreign matter, 80; se-
lectivity of, 65; standardization
of, 101, 109; temperature coef-
ficients of, 74

Dormancy at low temperature, 79
Drug-fastness by adaptation, 58
Dyes, 136; selective action of, 65

Enzymes, inactivation of, as cause
of death, 36 seq.; inactivation by
sulfonamides, 149; molecules
per cell, 37

Evaluation of antiseptics. 118, 120;
of disinfectants. 101, 109

Fatty acids, 98

Federal Drug Administration
method, 103

Flour beetle, death of, 14, 18

Formaldehyde, as antiseptic, 130;
as disinfectant, 90, 106, 112, 129
Fruitflies, death by heat, 18

Genes affected chemically, 41
Gentian violet, 127, 138
Gliotoxin, 165

Graded resistance, 25, 31, 35, 36,
45, 56

Gram stain, and antisepsis, 123;

and disinfection, 64
Gramicidine. 124, 165
Growth retardation by antiseptics,

150, 163

Halogens, 92

Hands, disinfection of. 96, 100

Hydrogen ions as disinfectants,

76, 82, 84

Hydrogen peroxide, 83, 89, 91, 106,
112

Inactivation, of enzymes, 36; of
genes, 40

180

SUBJECT INDEX

Inhibition, irreversible, 62; pei--
nianent, 62; reversible, 62; tem-
porary, 62

Inoculum, affecting antisepsis, 133,
143, and deathtime, 54

Iodine, 95

Katadyn, 89

Lag, extension by antiseptics, 162;
phase, prolonged by antiseptics,
130, 138, 143

Majority thermal death point, 46

Mass law and disinfection, 37

Mercuric chloride, 85, 86, 102, 113;
concentration exponent of, 67, 74

Mercurochrome, 87, 106

Mercury compounds, 85

Metals as disinfectants, 89

Metaphen, 87, 106

Methionine antagonizing sulfona-
mides, 156

Mold spores, death of, 50

Mortality curves, 13, 15, 22, 24, 26

Mosquitoes, death of, 14, 34

Multicellular vs. unicellular organ-
isms, 20, 24

Multiplication retarded by airti-
septics, 130

Mustard seeds, death by poison,
18, 34

Mutations, 41

Non-logarithmic order of death, 42

Olygodynamic action, 88

Organic matter, disturbing disin-
fection, 80. 84; measure of in-
fluence, 111

Oxidation-reduction potential, 139,
142

p-amino benzoic acid, 153
Penicillin, 165; causing morpho-
logical changes, 166; selectivity
of, 126
Permanganate, 91, 106, 112
pH affecting antisepsis, 140, 144,

1.54, 157
Phenol, as antiseptic, 130; coeffi-
cients. 101, 106, 111; concentra-
tion exponent of, 68, 73; disin-
fection by, 90, 106, 112

Physiological youth, 141
Propionic acid, 159
Protozoa, death of, 51, 52
Purines antagonizing sulfona-
mides, 156
Pyocyanin, 164, 165

Resistance, abnormal, of individ-
uals, 45, 46, 57
Resistant strains, 56
Rivanol, 137, 139

Salicylic acid, 158

Selection of resistant strains, 5G

Selective culture media, 128

Selectivity, of antiseptics, 65, 123;
of disinfectants, 63

Silver, colloidal, 88; nitrate, con-
centration exponent of, 69; salts,
88, 106. 112

Slow death, order of, 47

Soaps. 64, 66, 95

Sodium azide, selectivity of, 124

Specificity, see selectivity

Standardization of disinfectants,
64, 101, 109

Sterility, a lethal mutation, 41

Sulfathiazole, 139

Sulfonamides. 146; ionization of,
155

Sulfonated niacin, 154

Sulfur dioxide, 159, 163

Survivor curves, 13, 15, 24

Tadpoles, death of, 14, 21, 25, 34
Technique of evaluating disinfect-
ants, 101, 109, 116
Temperature, and antisepsis, 121;
and sulfonamides. 156; coeffi-
cients of disinfection. 7G; coef-
ficients of life and death, 78;
effect on disinfection, 74
Thickness of cell membrane, 35
Tcmato moth larvae, deathrate of,

34
Toxicity index, 109
Tribolium. death by X-rays, 14, 18
Tubercle bacteria, resistance of,
64

Wetting agents, 99

Wheat seeds, death by heat, 34

Yeasts, death of, 14, 15, 34, 48, 50

AUTHOR INDEX

Abraham, 126, 165
Abrams, 100
Albus, 42
Anderson, 93
Ardrey, 53, 93, 94

Baker, 100
Baldwin, 89, 91
Bambas, 149
Barnes, 50, 73
Barron, 149

Beamer, 14, 20, 36, 46, 50
Bechhold, 89
Belehradek, 79
Bell, 155
Bergmann, 89
Bigwood, 79
Birkhaug, 131, 132
Birkinshaw, 169
Birstein, 68, 72, 75
Bittenbender, 157
Borman, 58
Boycott, 14, 36
Bronfenbrenner, 109
Brooke, 116

Cameron, 122

Carroll, 57

Chaffee, 166

Chain, 126, 165

Chapin, 92

Charlton, 53, 75, 93, 95

Chick, 13, 18, 36, 43, 67, 72, 73 74

75, 76, 78, 81, 101, 102
Chu, 149
Churchman, 137
Clark, 18
Cohen, 75, 76
Conn, 129, 132, 157, 160
Cooper, 127
Coulthard, 169
Cruess, 122, 135, 157
Curran, 83

Daddi, 151
Dalby, 157

Daniels, 155

Davy, 18

Dawson, 165

Degering. 157

Demerec, 40

Diernhofer, 123

Dittmar, 89, 91

Doisy, 169

Dorfman, 150

Doubly, 109

Downs, 126

Dubos, 124, 138, 140, 144, 164, 165

Eagle, 168
Eggerth, 97, 98
Eijkman, 50
Ely, 109, 152
Epstein, 135, 137
Esty, 55
Evans, 83

Feasly, 157
Fildes, 153
Fletcher, 126, 165
Florey, 126, 165
Foley, 135, 137
Foter, 79, 152
Fox, 154
Fred, 57
Fricke, 40

Gaby, 169

Gage, 18, 46, 56

Gardner, 126, 165, 166

Garre, 164

Gates, 57

Gegenbauer, 26, 73, 74, 86, 120

Gilbert, 34, 36

Glover, 14, 34

Greig, 151

Groves, 34

Gwynn, 157

Haldane, 37

Hand, 37

Harris, 147, 148, 150, 156

182

AUTHOR INDEX

Harrison, 100
Hartmann, 123
Harvey, 51
Hastings, A. B., 149
Hastings, E. G., 57
Heatly, 126, 165
Henry, 146
Hershey, 109
Hewlett, IS, 34
Hirsch, 139, 150
Hobby, 165, 166
Hoerr, 99
Hoffman, 157
Hoffmann, 138, 141-144
Holwerda, 92
Homan, 169
Hoogerheide, 151
Hotchkiss, 124
Hoiick, 166
Hiiber, 100

Ingle, 159

Ingraham, 127, 138, 142
Irish, 157
Isaacs, 36
Jacobs, H. R., 149
Jacobs, M. H., 161
Jennings, 126, 165
Johnson, 147, 154
Jones, 146, 169

Kain, 169
Katzin, 30
Katzman, 169
Klarman, 96, 97, 98
Klein, 100
Knaysi, 31, 47
Kocholaty, 169
Kohn, 147, 148, 150, 156
Krampitz, 168
Krause, 89
Kronig, 66, 69, 84, 85
Kulman, 169
Kumler, 155

Lamanna, 156
Landy, 153
Lange, 58
Larkum, 153

Lee, R. E., 34, 36
Lee, S. W., 135, 137
Leonard, 88
Levine, 53, 75, 93, 95
Leviton, S3
Lloyd, 34
Loeb, IS, 31, 34
Lommel, 41

MacLeod, 133, 151

Macy, 160

Madsen, 13, 34, 75

Magoon, 57

Mallmann, 53, 93, 94

Martin, 81, 101, 102

Mason, 127

McCullough, 64, 76, 78, 84, 106, 122

Mcllwain, 154

Mcintosh, 152

McKee, 166

Meader, 58

Mellon, 149

Meyer, 55, 165, 166

Michaelis, 169

Miller, B. F., 100

Miller, S. B., 89, 91

Mirik, 133

Mitchell, 156

Moore, 147

Muir, 146, 169

Miiller, 74

Murray, 167

Myers, 34, 85

Northrup, IS, 31, 34
Nyman, 13, 75

O'Kane, 14, 34
Olson, 160
Oswald, 153

Paul, 38, 66, 68, 69, 72, 75, 84, 85
Penfold, 41
Peters, 51
Phelps, 101, 102

Rahn, 20, 40, 42, 57, 79, 138, 142,

148
Raistrick, 169
Ralston, 99

AUTHOR INDEX

183

Reichel, 73, 90
Reichenbach, 34, 35, 45
Reithel, 169
Rettger, 142
Reuss, 68, 72, 75
Revis, 41
Rice, 150

Richert, 122, 135, 157
Rideal, 101
Roberts, 169
Roblin, 155
Rogers, 161
Rose, 154

Sandholzer, 30
Sattler, 34
Saunders, 150
Schaffer, 97
Schmidt, 166
Scliweitzer, 157
Sesler, 166
Sevag, 151
Shamleffer, 146
Shapiro, 156
Shelbiirne, 151
Sherman, 42
Short, 169
Shternow, 96, 97
Skrimshire, 169
Slanetz, 142
Smelkes, 152
Smith, J. H., 35, 50, 75
Snell, 156
Standfast, 169

Starkey, 85
Stearn, 138, 145
Steinberg, 41
Strandkov, 152
Sykes, 169

Tanner, 14, 20, 36, 46, 50

Tetrault, 157

Thorn, 41

Thomas, 106

Tilley, 71-76, 89, 92, 97

Tunicliff, 167

van Bruggen, 169

Waksman, 85, 165
Walker, J. E., 96
Walker, J. T. A., 167
Watkins, 18, 20, 33, 59, 108
Watson, 67, 70, 72, 101
Welshimer, 168
Werkman, 168
Westgate, 14, 34
Whitby, 152
White, 156
Whittier, 161

Winslow, 20, 33, 34, 59, 116
Woerz, 47, 50
Woods, 153
Woodruff, 165
Wyckoff, 14, 42, 44
Wyss, 152

Zeug, 110

CLYNN PRINTINC: COMPANY

SAINT I.OI.IS, MISSOIKI