Infection & nutrition



Edward Mellanby M.D., F.R.C.P., F.R.S.

Late Professor of Pharmacology, University of Sheffield
Consulting Physician, Royal Infirmary, Sheffield
Secretary of Medical Research Council

Puerperal Sepsis
Experimental Production of Puerperal Sepsis
Puerperal Septicaemia
Dietetic Factors responsible for Perfect and Imperfect Teeth
The Structure of Human Teeth
Relation of Dental Structure to Caries
The Defence of Teeth
The aetiology of Periodontal Disease (Pyorrhoea)


there is general agreement among medical men that the susceptibility of mankind to many types of infection is closely related to the state of nutrition. The difficulty arises, when closer consideration is given to this general proposition, as to what con­stitutes good and bad nutrition, and the problem is not rendered easier by recent advances in knowledge in nutritional science. Even taking the older criteria of malnutrition—height and weight—these may not indicate any defect in a person who is naturally small and thin. On the other hand, using more recent knowledge as to the effects of specific deficiencies or excesses in the diet, these may be too exacting as tests of malnutrition. Thus a well-grown, healthy looking child with badly-formed and carious teeth ought to be regarded as suffering or having suffered from malnutrition. The bones of such a child are probably also defectively calcified, but this may not be revealed by ordinary examination but only by radiographic examination or by the greater ease with which these bones may be fractured. Similarly, a child with a goitre but otherwise well-grown and healthy is passing through a period of specific mal­nutrition. The same applies to anaemia. If we take these tests as criteria, a large part of the population of this country is suffering from malnutrition.

There are, however, some types of malnutrition which are related to serious illness. One of these is the result of a deficiency in the diet of vitamin C, Which brings about scurvy; another, the sequelae of a vitamin A deficiency, which I propose to discuss now. Although we have fairly clear ideas as to the clinical indications of an absolute deficiency of vitamins C and A, yet our knowledge as to the effects of partial deficiency of these substances is very indefinite.

At the present time, therefore, we are at an impasse on this problem of malnutrition. We know too much to take the older views as to the criteria of malnutrition seriously, and we know too little to lay down specific rules as to what criteria should be used according to recent knowledge. One thing seems certain, that in future new standards with new instruments of precision will become of greater importance in determining this condition. Clearly the only way out of the difficulty is to accumulate more accurate data showing the relation of dietetic abnormalities to disease, and in the present lecture I propose to consider from this point of view the problem of the susceptibility of animals and man to certain types of infection. It must be realised that this subject is only in its infancy, and it is impossible to foresee how it will be extended by future work. My own belief is that the relation of nutrition to infection will ultimately prove to be a subject of great practical importance, although the facts so far estab­lished may require a different interpretation.

The laboratory played the initial part in this particular problem of the relation of diet to infection. Many observers have drawn attention to the fact that when a diet is deficient in fat-soluble vitamins, animals are very susceptible to infection (M'Collum,1 Drummond,2 Mellanby,3 Cramer and Kingsbury,4 Steenbock, Sell and Buell5).  Xerophthalmia in particular was regarded as due to a specific deficiency of a fat-soluble substance (Osborne and Mendel, 6 M'Collum and Simmonds7), but the advance in the study of the problem was held up for some years because it was not realised that fat-soluble A was usually a complex of two vitamins, one of which retained the name vitamin A and the other, the anti-rachitic vitamin, was called vitamin D. However, before these vitamins were differentiated and separated, Mori8 described the changes in epithelia characterised by hyperplasia as due to a fat-soluble vitamin deficiency, the study of which was afterwards developed by Wolbach and Howe.9 Goldblatt and Benischek10 subsequently proved that these changes were due to absence of vitamin A (as we know it to-day), for they produced the hyperplasia in animals even when vitamin D was included in the diet. For a particular case of epithelial hyperplasia due to vitamin A deficiency, see Figs. 22 and 23.

Prior to 1926,11 I became interested in the problem of infection owing to an outbreak of broncho-pneumonia among my experimental dogs.  On analysing the diets of these animals, I found that susceptibility to infection was independent of the condition of the bones, that is to say, independent of the vitamin D intake. This can be seen in Table XI.

Atelectasis in these cases was associated with rickets and was often present when vitamin D was very deficient in the diet. On the other hand, it will be noticed that broncho-pneumonia never developed when the diet contained a good source of vitamin A (for instance, when butter was given), even though the lungs might be collapsed. It may be added that the chief reason why rickets developed and


Lung Condition found at Post-Mortem Examination in a Series of 330 Dogs in Relation to Diet


Condition of Lungs.

Diet as regards Vitamins A & D


Vegetable fat.
Vitamins A & D Deficient

Butter. Vitamin A+, Vitamin D some

Butter heated and oxygenated.
Vitamin A-,
Vitamin D-.

Cod-liver Oil.

Vitamin A+,
Vitamin D+.

Cod-liver Oil heated and oxygenated.
Vitamin A-,
Vitamin D some







Broncho pneumonia


















the lungs collapsed in patches when butter formed part of the diet is that oatmeal was the cereal of the diet in many cases, and, when this is so, butter is of comparatively little value as an anti-rachitic agent, unless the calcium of the diet is sufficiently high (see pp. 9, 16, 17, and Fig. 5).

The deduction from the above observations is that vitamin A and not vitamin D protected the lungs from inflammatory changes under the particular conditions.

These results suggested a search for the cause of the death of rats on a diet deficient only in vitamin A.

Goldblatt and Beneschek, as mentioned above, had found that this deficiency resulted in hyperplasia and local infection, but they were more interested in the significance of the hyperplasia and the metaplasia than in that of the infection. They regarded these local infections as an episode and not as a certain cause of death, as our subsequent experiments showed was the case.  In a large-scale experiment on this point. Green and Mellanby12 found infective foci almost without exception in young rats brought up on diets which were complete except for vitamin A. The distribution of the lesions in 92 such animals is given in the following table :— 

table XII.

 Distribution of Infective Foci in Rats brought up on a Vitamin A-Deficient Diet 


Infection of genit-urinary tract, including pyelo­nephritis, cystitis, renal and bladder calculi, hydronephrosis, dilated ureters and distended bladder   



Middle-car disease and septic nasal sinuses


Acute inflammation of small or large intestine (acute enteritis involving duodenum and jejunum, often immediate cause of death), also recent haemorrhage into pyloric end of stomach


Infection of lungs




Abscess in floor of mouth, with suppurating glands in neck   

72 up to 90 with older rats

 The above were the main kinds of infective lesions found, but abscesses in many other places occurred and often it was possible to grow micro-organisms directly from the blood. If a source of vitamin A, such as butter, cod-liver oil or egg yolk formed a part of the diet, infective lesions were never seen in the rats and the addition of these substances to the deficient diets, unless the animals were too severely infected, generally resulted in rapid improvement and ultimate cure.

Following V. Euler's13 observations on the re­placement of vitamin A by carotene for growth of rats, Green and Mellanby14 showed that carotene also prevented infection, and indeed with quantitative alteration in the amount of carotene used, the response in intensity of infection was very well graded, as the following results show:—


Degree of Anti-infective Action of Various Doses of Carotene given Daily to Rats on a Vitamin A-deficient Diet

Protective agent

Severe Infection

Moderate Infection

No Infection





0.005       mg. carotene




0.010               “




0.020              “




0.400              “












0.500 g. dried cabbage




 In the above table we see that with 5 y, i.e. 5 one-millionths of a gram of carotene, there is very little, if any, protection against infection, while with 40 y protection is complete. It is probable that, if this work were repeated with the purer samples of carotene now available, protection against infective lesions would be procured by smaller doses than the above.

It will be further noticed that 500 milligrams of dried cabbage, presumably because of its carotene content, always gave complete protection.

Thus vitamin A and carotene under these defined laboratory conditions act in a specific way against widespread local infective lesions. Because of this striking effect, Green and Mellanby16 called vitamin A the anti-infective vitamin in order to draw attention to what seemed to be its most characteristic action. We were, of course, aware of the drawbacks of giving a label of this kind, because the word "infection" covers a vast number of different types of pathological phenomena, but we also recognised that it had the advantage of attracting the attention of workers to this important subject. Before passing on to the clinical side of the problem, it may be well to say that the results on rats have been confirmed by other workers (Turner and Loew15) and appear to be firmly established. They are so clear-cut and definite and the same types of septic foci are so commonly found in human beings that the immediate deduction made was that here we have a phenomenon of direct clinical interest, and there is no wonder that we soon made an attempt to test this supposition on patients. There were, however, two reasons why we had some doubt as to the direct applicability to human infection—(1) the very ease of producing, preventing and curing the septic foci in experimental animals, and (2) the difficulty in believing that diets even of poor people were as deficient in vitamin A and carotene as the experimental diets. It was, however, clear that poor people, whose diets tended to be deficient in the more costly foods supplying vitamin A, i.e. milk, butter, eggs and green vegetables, suffered more frequently from the types of infection described than those more wealthy, and we thought it worth while to see whether vitamin A played a part, even if not so great a part, as in the animal experiments, in controlling some kinds of human infection. We therefore turned our attention to the clinical aspect of the case and made two investigations, one on puerperal sepsis, the second on puerperal septicaemia.

Puerperal Sepsis

In this test, 550 women attending the ante-natal clinics in Sheffield were investigated (Green, Pindar, Davis and Mellanby16). Alternate women (275) were given a supply of a preparation rich in vitamins A and D during the last month of pregnancy, the remaining women were not given any supplement. No vitamin preparation was given to any patient after entering hospital.  No instructions were given as to diet in any case. The women at full term were brought into hospital and delivered, the attend­ing doctors having no idea of previous therapy. After discharge from the hospital, all the notes of the patients were collected and analysed. The results were as follows:

The morbidity rate in the puerperium using the B.M.A. standard was n per cent. in the vitamin group and 4-7 in the control group, a difference of 3-6 per cent. which is twice the standard error (1-4), and therefore statistically significant.

 At one hospital where over 400 of the cases were delivered, the morbidity rate in the treated group was l.0 per cent. as compared with 5.8 per cent. in the untreated group. These rates compared with 7-3, 8.1 and 5.0 per cent. in the total ante-natal cases delivered in this hospital in the three years prior to the commencement of this investigation.

When the morbid cases were classified, not the B.M.A. standard, but on the notifiable standard (Public Health Regulations), the difference in the morbidity rates between the two groups was not so great, the rate being 4.0 per cent. in the vitamin group and 5.5 per cent. in the controls. This means that later in the puerperium, i.e., after the first ten days, possibly owing to the using up of the vitamin reserves (vitamin therapy stopped .on admission) some of the vitamin group developed sepsis. How­ever, even so the difference between these morbid cases in the two groups was clinically very great; thus 12 cases were classified as clinically severe in the vitamin group and 26 in the control group.

The result of this large investigation suggests that vitamin A therapy given before childbirth has increased the resistance of the genito-urinary tract to invasion by micro-organisms.

Experimental Production of Puerperal Sepsis

Let us now turn back to the laboratory and see whether this clinical experience receives direct support from the experimental side. The following investiga­tion was made by H. N. Green in my laboratory.

Three groups of does (rats) were placed on a standard basal diet which was vitamin A deficient (Green and Mellanby, 1928). When pregnant each rat was separated and maintained on the group diet until delivered-; she was then returned to the communal cage.  In group 1. only 5 out of 13 became pregnant and had litters which were in each case either born dead or were eaten soon after birth. All 13 does died of infection. In the 8 non-pregnant infection was -localised mainly In the lungs. In the 5 which delivered litters the genito-urinary organs

diet to 9 does. In every case the rat became preg­nant, more than one litter being born in most cases. Of these animals, on post-mortem examination, 8 of the 9 were found to be free from infection of the generative organs; one showed slight infection. Three developed septic foci in other organs.   Although many litters were born from these rats, scarcely any of the young survived. Since we know from previous experience that litters can be successfully reared on this basal diet if adequate amounts of cabbage are given, it is probable that the 2 g. of moist cabbage given daily were insufficient in regard to the carotene content. (Other rats which were given cabbage ad lib. had several litters; the young lived; and later, when killed, the mothers were found to be normal.) The vitamin A content of the livers of this group was found to be relatively low (average figure being 32 blue units per g. of liver). This finding illustrates the point of the necessity for an increased vitamin A intake during pregnancy, for in the non-pregnant; rat 2 g. of green cabbage taken daily is ample for main­taining health.

These experimental results show that the mucosa of a parous uterus is more susceptible to infection when the diet is relatively deficient in vitamin A, during or after pregnancy, and are in keeping with the clinical results in women above described. Taken in conjunction with the results of the clinical investiga­tion they leave but little doubt that this nutritional aspect of preventing puerperal sepsis in women is one of practical importance. Following pregnancy, women are peculiarly susceptible to infection, and the evidence suggests that the great attention given in recent years to the bacterial aspect of sepsis in parturition has by no means solved the problem. In an analysis of women suffering from puerperal sepsis, Rivett, Williams, Colebrook and Fry17 found that serious sepsis appeared in a large number follow­ing normal spontaneous labour, without perineal or vaginal lacerations and without vaginal examination having been made. Facts such as these and the knowledge that pregnancy and lactation involve not only a metabolic upset, but often a sacrifice of essential nutritional substances to the foetus, emphasise the probable importance of this aspect of the subject,

Puerperal Septicaemia

Before reviewing our clinical investigation on puerperal septicemia, it may be well to point out that not only is this problem fundamentally different in type from that of puerperal sepsis, but our method of investigation was also different (Mellanby and Green18), In the work on puerperal sepsis we were testing the action of a fat-soluble vitamin preparation prophylactically in respect of its power to raise the resistance of a mucous membrane. In puerperal septicaemia we were testing the curative effect of diets on the resistance of the body to micro-organisms after invasion of the blood-stream. Our experimental results on animals had not suggested that vitamin A increased the general resistance to infection of this type.   They did show, however, that mucous membrane resistance was increased by the presence of the vitamin.  Again, whereas in the puerperal sepsis investigation, we had used only a fat-soluble vitamin preparation as the test substance, controlling each case receiving the supplement with a case not receiving it, in the puerperal septicaemia investigation all patients received when possible a diet rich not only in vitamin A but also of high biological quality. This diet included much milk, eggs, green vegetables; etc., as well as the vitamin A supplement. For controls we had to use the cases treated in previous years by the same obstetricians and gynecologists as the test cases.

Generally speaking, then, this investigation is not so scientifically sound as that on puerperal sepsis, but it seemed to us that if we could get some idea as to the value of good nutrition on raising the resistance of the body to puerperal septicaemia we could after­wards simplify the problem by more rigid tests.

Altogether 37 cases of puerperal septicaemia in which growths of micro-organisms were obtained from the blood were treated and the results obtained ire classified as follows:


Result of Treating Cases of Puerperal septicaemia with Diets of High Biological Value Rich in Fat-Soluble Vitamins 

Organism isolated from Blood. 

No of cases

No of deaths

Mortality rate

Streptococcus haemolyticus



28 %

B. coli






B. paratyphosus



Total cases




 It may be added that at the same hospital in the years 1937, 1928, and the early part of 1929 prior to the above investigation, the mortality rate in 18 consecutive cases of haemolytic streptococcal septicaemia was 92 per cent. The medical attendants during these periods and the pathologist investigating the blood growths were the same. At the same time, the control cases and the diet investigated cases are not closely comparable and all that can be claimed is that the effect of the diet seems to have been associated with a mortality rate much lower than is usually found.

Our general impression of the dietetic treatment is that in more or less straightforward cases of puerperal septicaemia, the effect was distinctly beneficial. The complications that made the treatment in some cases ineffective were (i) peritonitis, and (2) extensive thrombo-phlebitis. Nor is the treatment of value in the extreme fulminating type of infection. Complica­tions such as pneumonia and the development of localised abscesses were not regarded as so serious.

The development of extensive thrombo-phlebitis and its high mortality, had one aspect which may be significant.  Green19 examined the vitamin A content of the livers of these patients and found that often even after intensive vitamin A treatment, there was only a trace of the vitamin present in contrast to the normal animal which usually stores a part, and often a large part, of the vitamin A given. This inability to use vitamin A in the normal way may be related to the ineffectiveness of vitamin A therapy in cases of septicaemia associated with extensive thrombo-phlebitis.

I have given this outline of the investigation on puerperal septicemia, not because it proves the point at issue, viz. the anti-infective action of vitamin A, but because it raises the whole question of increasing the resistance of the body to micro-organisms in the circulatory system by nutritional influences. It seems to me to be a line of inquiry worth further considera­tion; attention ought to be directed to the whole biological qualities of the diet as well as its vitamin A content.

In conclusion, I may say that I have tested the effect of a vitamin A rich diet, which included much milk, egg yolk and vegetables on several cases of malignant endocarditis, but the results were dis­appointing. Although the general condition of the patients improved, the valvular lesions did not heal up, nor were embolic sequelae diminished.


One of the most interesting tests of the value of vitamin A in infective diseases is that of Ellison,20 on the treatment of measles. This disease was chosen by him because of the involvement of the respiratory epithelium which is so prominent a feature. It has already been pointed out that experimental results had indicated that vitamin A was particularly concerned in the defence of epithelial surfaces, including that of the respiratory tract. Ellison tested the effect of a vitamin concentrate rich in vitamins A and D on 300 cases of measles under the age of 6, using another 300 cases of similar age as controls. His general results were:—

(1) The mortality in 300 vitamin treated cases was 3.7 per cent.
(2) The mortality in 300 control cases was 8.6 per cent.
(3) The vitamin treatment lowered the mortality of cases with pulmonary complications.
(4) Vitamin therapy did not lower the otological and cutaneous complications.

The following are his figures showing the effect of the vitamin therapy on pneumonia complicating measles:


Effect of Fat-Soluble Vitamins on Pneumonia complicating Measles (Ellison) 



Vitamin Cases


Cases of pneumonia


Cases of pneumonia




























34 (5)

23 (5)

32 (2)

10 (2)

Figures in brackets represent cases developing pneumonia after admission to hospital.

 These results, like those described above concerning the prophylactic tests in puerperal sepsis in women, show that in some infective disorders of mucous membranes affecting human beings, vitamin A raises the resistance and diminishes the incidence and severity of such disorders. It is of interest to note that, according to Ellison, this form of treatment in measles did not diminish the incidence of middle-ear disease, a condition which in the experimental animals was directly under the control of vitamin A and carotene. This result is in keeping with the fact that I have seen middle-ear disease of an acute type develop in children where the vitamin A reserves were certainly high.

The difference between the conditions under which infective lesions in the experimental animals and those under which puerperal sepsis and measles in human beings develop in human beings is that in the former case vitamin A and carotene were absent or almost completely absent from the diet, whereas in the clinical cases, although probably small, there is no reason to believe that the deficiency was so great. It is indeed clear that infective foci of the type seen in vitamin A deficient animals develop in human beings independently of the vitamin A reserves and therefore may have a different etiological significance. On the other hand, there is good evidence that many of the experimental lesions found in animals do actually develop in human beings when the vitamin A intake is known to be low. Thus Bloch21 examined the incidence of infective diseases in 86 patients in whom deficiency of this vitamin was confirmed by the presence of xeroph-thalmia. In these 86 cases, 68 suffered from some infective disease, including 15 cases of pneumonia, 12 of bronchitis, 15 of otitis media, 27 of pyuria and 14 of pyodermia. Results like these are in complete harmony with the experimental findings in animals. The increased susceptibility of man to infections of the skin in vitamin A deficiency have been noticed by others than Bloch.43 References to such lesions in association with keratomalacia were made by Mori,22 Pillat,23 Spence,24 and Loewenthal.25 This subject is of interest in relation to the evidence (p. 129) that vitamin A deficiency may play a part in the aetiology of pellagra.

Corroboration of these results when vitamin A and carotene is deficient in the diet can be seen in the high incidence of infective disorders found in the prisoners at Kampala in Uganda by Mitchell.44 In 1937 when xerophthalmia with keratomalacia was common among these prisoners, dysentery, diarrhoea, pneumonia and bronchial affections were rife and the death-rate abnormally high, i.e., up to 200 per 1000 per annum. By altering the diets and other means this high death-rate was brought down to 14 per 1000. The high incidence of bronchitis, pneumonia, tropical ulcers and phthisis among the Kikuyu tribe who live on a diet mainly of cereals as compared with the low incidence of these diseases among their neigh­bours the Masai who live on meat, milk and raw blood (Orr and Gilks"), probably has a similar or related nutritional explanation. The differences in distribution of infective disease found by these workers in the two tribes are most impressive. Thus in the cereal-eating tribe, bronchitis and pneumonia accounted for 31 per cent. of all cases of sickness, tropical ulcers for 33 per cent. and phthisis for 6 per cent. The corresponding figures for the meat, milk and raw blood tribe were 4 per cent., 3 per cent. and l per cent.

Clinical evidence therefore shows:—

(1) That human beings whose diet is almost devoid of vitamin A or carotene often develop septic lesions as found in animals under the same circumstances. The mortality in these cases is high and the situation can be remedied by vitamin therapy.

(2) That, when the vitamin A reserves are not depleted, but when nevertheless there is reason to believe that some deficiency may exist, human beings are more susceptible to certain types of infection and their resistance to these infections can be raised by vitamin feeding.

(3) That local infective conditions can develop in man even when the vitamin A reserves are, so far as we know, adequate.

'The main difficulty to be solved centres round the second of these conclusions. Is it possible to prevent or cure infective conditions, apart from those already described, by high vitamin A or other nutritional treatment? There is some evidence that tubercular infection can be influenced by fat-soluble vitamins. The extensive use of cod-liver oil and its generally accepted therapeutic value in tuberculosis, supports this view. Statistics on this point are, however, difficult to procure. M'Conkey27 found, in cases of intestinal tuberculosis complicating pulmonary disease, that, whereas the mortality of untreated cases (28 in number) was 71 per cent., the mortality among 50 cases who had received additional cod-liver oil and tomato juice was only 10 per cent. The mortality in 50 cases treated by heliotherapy was 24 per cent., so he attributes much of the curative action to vitamin D, although experimental results on animals do not support this explanation.

Many laboratory investigations have been made to gain further knowledge of the relation of nutritional conditions on susceptibility to infection. In my own laboratory, H. N. Green29 has made an extensive study, and others, including Lassen,28 Boynton and Bradford,30 Zilva,31 Ruth Green,32 have investigated the problem. On the whole these studies have not added much to our knowledge. One point of interest that may open up the problem again is the discovery (described on p. 112) that xerophthalmia is related to degeneration of the sensory nerve supplying the cornea, i.e., the trigeminal, and it is possible that other infective foci in vitamin A deficiency may also be related to loss of neurotrophic control, as indicated by demyelination changes in the afferent nerves supply­ing these organs.

It is clear from this brief review of the situation that much remains to be done in order to establish in a more precise way and to extend this relationship between nutritional defect and susceptibility to many infective conditions. There is no doubt, however, that a diet devoid of vitamin A results in a great diminution in resistance to infection in human beings. One rich in fat-soluble vitamins and of high biological value increases the resistance to infection. The gap between these two positions wants filling up and demands investigation both experimentally and clinically.

It is probable that, as in the case of vitamin D and rickets, the question is not simple and that it will ultimately be found that vitamin A works in harmony with some dietetic factors, such as milk proteins and other proteins of high biological value, to promote resistance of mucous membranes and epithelial cells to invasion by micro-organisms, while other factors such as cereals, antagonise its influence. The effect of increasing the green vegetable and reducing the cereal intake on the resistance of herbi­vorous animals to infection is undoubted (Glenny and Allen,33 Boock and Trevan34) and may well indicate a reaction in which the increased carotene of the vegetable plays only a part, but an important part. Similarly, the results obtained by Orr, Macleod and Mackie35 on the effect of nutritional changes on the serological reactions of sheep indicate a definite but possibly complicated relationship between diet and susceptibility to infection.

It may be objected that in this discussion I have laid too much stress on the positive experimental and clinical results, and have not dwelt sufficiently on those clinical investigations made since we described vitamin A as the anti-infective vitamin and which have given negative results. Investigations in which vitamin A therapy has proved to be ineffective include the following: (1) Barenberg and Lewis 36 on respiratory infections in infants; (2) Sutliff, Place and Segool 37 on otitis media complicating scarlet fever; (3) Wright, Frosst, Puchel and Lawrence38 on the common cold in infants; (4) Orenstein39 on pneumonia. In all these investigations the control cases were receiving some vitamin A in their diets, but the test cases were getting further supplements of this vitamin. I am quite aware of the possibility of this criticism, but in excuse I may frankly state that my object is to force open this closed door behind which, I believe, is the secret of the correlation of nutrition and susceptibility to infection, and with this object in view, I am prepared to use every available weapon which promises to lead to further knowledge. On account of the positive evidence described, negative results only show that our knowledge is still very imperfect, but not that the problem is an unprofitable one to pursue further.

Vitamin A deficiency certainly lowers the body resistance to some types of infective invasion in man and animals. Equally, the same infective invasions certainly take place in the presence of vitamin A. The question is: Is there a common basis which may explain these two conditions?  Is there, for instance, some abnormality or loss of neurotrophic control which is a common factor in both cases?  This is only one of many possibilities.   In any case the most promising method in this and any other researches is to work from the known, established by experiment, to the unknown, and we do now know how to produce without fail many infective lesions experimentally in animals by removing vitamin A from the diet. On this foundation future investigators of this important problem can easily build.

In recent years, evidence has come from another quarter of the influence of diet on resistance to infection. I refer to the observations of Helmholz40 and Clark 41 on the curative effect of a ketogenic diet in many cases of urinary infection. It is true that the effectiveness of this therapy is regarded as depending on the actual excretion of ß-oxybutyric acid (Fuller42), and the acid reaction of the urine, but it would be surprising to me if these cereal-free and high vitamin A-containing diets commonly given to such patients had not the power of raising the resistance of the body to infection, apart from the ketogenesis they produce. Evidence of this nature ought to be examined and the use of such diets extended to other forms of infection than those of the urinary tract. The curative effect of diets devoid of cereal and rich in fat-soluble vitamins on dental caries (M. Mellanby and Pattison) has already been referred to (p. 25). It may prove that the best form of diet for combating sepsis is a combination of high vitamin qualities, especially of high fat-soluble vitamin content, with low cereal and a definite ketogenic action.


1. M'Collum, E. V.. .   . J. Amer. Med. Ass., 1917, 68, 1379.
2. Drummond, J. C. .   . Biochem.J., 1919,13, 95.
3. Mellanby, E. .   .   . Lancet, 1919,1, 407; and Sp.Rep. Ser. Med. Res. Coun. Lond., No. 61, 1921.
4. Cramer, W., and Kings'bury, A. N, .   .   . Brit.J. Exp. Path., 1924, 5, 300
5. Steenbock,   H.,   Sell, M. T., and Buell, M. V. J. Biol. Chem., 1921, 47, 89.
6. Osborne, T. B., and Mendel, L. B.   .   .J Biol. Chem., 1913, 16, 423.
7. M'Collum, E. V., and Simmonds, N.   .   J Biol. Chem., 1917, 32, 181.
8. Mori, S. .   .   .   .
Bull. Johns Hopkins Hosp; 1922, 33, 357.
9. Wolbach, S. B., and Howe, P. R..   .   . J. Exp. Med., 1925, 42, 753.
10. Goldblatt, H., and Benischek, M.    .   .   . J. Exp. Med; 1927, 46, 699.
11. Mellanby, E. .   .   . Brit. Med.J., 1926, 1, 515.
12. Green, H. N., and Mellanby, E.    .   . Brit. Med.J.. 1928,2, 691.
13. Euler, B. von, Euler, H.von, and Hellstrom, H. Biochem. Z., 1928, 203, 370.
14. Green, H. N., and Mellanby, E.    .   . Brit.J. Exp. Path., 1930, 11, 81
15. Turner, R. G., and Loew, E. R.   .   .   .   J Infect. Dis; 1933, 52, 102.

6. Green, H. N., Pindar, D., Davis, G., and Mellanby, E.  .   .   . Brit. Med.J., 1931,2,595
17. Rivett, L. C., Williams, L., Colebrook, L., and Fry, R. M. .   .   . Proc. R. Soc. Med., 1933,26, 45. (sec of Obstet. and Gynaecol.)
18. Mellanby, E., and Green, H. N. .   .brit med J, 1929,1,984
19. Green, H. N. . ..   . Lancet, 1932, 2, 723.

. Ellison, J. B     Brit. Med.J., 1932, 2, 708.
21. Bloch, C. E.   . Acta pediat; 1928, 7,61.
22. Mori, M. . Jahrsber. Kinderkl., 1904, 59, 175.
23. Pillat, A. . China Med J 1929, 43, 907.
24. Spence, J. C. Arch Dis. Child., 1931, 6, 17.
25. Loewenthal, L.J.A  African Med., 1933, l0, 58.
26. Orr, J.B..and Gilks,J.L. . Sp. Rep. Ser. Med. Res. Coun No. 155, 1931
. M'Conkey, M.        Amer. Rev. Tuberc., 1930, 21, 627.

28. Lassen, H. C. A      J. Hyg., 1930, 30, 300.
29. Green, H. N. .       . Proc. R. Soc. Med, 1933, 26, 34. (Sec of Comp. Med.)

30. Boynton, L. C., and Bradford, W. L.:    J Nutrition, 1931, 4, 323.   

31. Zilva, S. S.    .   .   Lister  Institute Ann1931 Rep (London).

32. Green, R. M. .   .   . Amer.J. Hyg., 1933,17, 59.
33. Glenny, A. T., and Allen, K. ....Lancet, 1921, 2, 1109.   
34. Boock, E., and Trevan, J. Biochem.J., 1922, 16, 781
35. Orr, J. B., Macleod, J. J. R., and Mackie, T. J. . Lancet, 1931, 1, 1177.
36. Barenberg, L. H., and Lewis, J. M..   .   . J Amer. Med. Ass., 1932; 98, 199.
37. Sutliff, W, D., Place, E. H.,and Segool, S. H.   J. Amer. Med. Ass., 1933, 100, 725

38. Wright, H. P.. Fr. Frosst, J. B., Puchel, F., and
Lawrence, M. R.     . Canadian Med. Assn. Journal, 1931, 25, 412.
39. Orenstein, A. J.        . S. African Med. J., 1932,6, 685.
40. Helmholz, H. F       . Mayo Clinic Proc., 1931, 6, 609;. Amer. Med. Ass., 1932, 99, 1305.
41. Clark, A. L           . Mayo Clinic Proc., 1931, 6, 605;
Lancet, 1932, 2, 511.
42. Fuller, A. TA          Lancet, 1933,2, 855.   
43. Bloch, C. E.          
Amer.J. Dis. Child..1931,42,263
44. Mitchell, J. P         East African Med.J., 1933, 10,38


in some ways the example of recent investigations on dental structure and disease illustrates even better the advantages of combined laboratory and clinical study, for the development of the story I am about to relate has remained in the hands of one person— M. Mellanby—from the time of its initiation by her. Whereas the rickets work began as an ad hoc in­vestigation, the study of the dental problem began in a chance observation made by M. Mellanby in 1917 on rickety dogs. She noticed that the structure of dogs’ teeth was altered by small changes in the diet, and realising its possible significance began one of the most persistent and intensive investigations that this country has ever seen 1-10.

Dietetic Factors responsible for Perfect and Imperfect Teeth

The first stage of the work was to determine the factors responsible for the structure of teeth.1-4 Since these, like bones, are largely composed of calcium phosphate the metabolic processes controlling their structure are in many ways similar to those influencing bone architecture. For purposes of study they have an advantage over bones in that, once formed, apart from absorptive processes preceding shedding of the deciduous teeth, their texture does not change like that of bones, so that a careful microscopic examination of fully developed teeth gives a much truer history than do the bones of the metabolic changes to which they have been subjected during the developmental period. It is now possible to produce at will in animals teeth of all grades of structure—from perfect texture to the greatest degree of imperfection—by making small variations in the food ingested. Thus, if growing puppies are given a limited amount of separated milk together with cereals, lean meat, orange juice, and yeast (i.e., a diet containing sufficient energy value and also sufficient proteins, carbohydrates, vitamins B and C, and salts), defectively formed teeth will result. If some rich source of vitamin D be added, such as cod-liver oil or egg-yolk, the structure of the teeth will be greatly improved, while the addition of oils such as olive or arachis oil leaves the teeth as badly formed as when the basal diet only is given (Fig. 11). Not only are the defects in the dentine and enamel obvious on microscopic examination, but external examination reveals the surface enamel also to be badly formed (Fig. 12). If, when the vitamin D intake is deficient, the cereal part of the diet is increased, or if wheat germ replaces part of the white flour (Fig. 13 (a)), or, again, if oatmeal is sub­stituted for white flour (Fig. 13 (e)), then the teeth tend to be worse in structure, but if, under these conditions, the calcium intake is increased, then calcification is improved (Fig. 14). Calcium appears both to antag­onise the anticalcifying effect of cereals and to aid the action of any vitamin D present in the diet, and this action becomes of great importance if butter is the fat of the diet. Butter in itself has a comparatively small calcifying influence in the presence of oatmeal, but a corresponding quantity of fat given as milk instead of butter, or an additional amount of calcium with the butter, may result in the development of perfectly calcified teeth (Fig. 15 (a), (b) and (c)).

It is obvious, therefore, that, although vitamin D holds the key-position, dental structure and incidentally bone formation cannot in actual practice be regarded as controlled by only one factor, but that these developmental processes are complicated mechanisms controlled by a series of interactions, some favouring and some antagonising perfect archi­tecture. In addition, it may be stated that, since the earlier stages of calcification are more unstable than the later ones, it is most important to get the beneficial influences to work as early as possible; and this can be accomplished only by suitably feeding the mother during pregnancy and the offspring in its early months of life.

If two comparable bitches are fed, one (A) on a diet of high calcifying qualities, the second (B) on a poor calcifying diet, the teeth of the offspring are affected in two ways :—

(l) The actual calcification of the teeth taking place in utero of the foetuses of A is better than in the foetuses of B.

(a) After birth, if the diets of the puppies of A and B respectively are of low calcifying qualities, the teeth of A's offspring—the well-fed mother—stand up more effectively to the bad conditions and are better formed than those of B's offspring. In other words, it would appear that if once the mechanism of calcification gets a good start as the result of perfect conditions in utero it is more difficult to upset by subsequent bad conditions.

The Structure of Human Teeth

The next stage of this investigation was to see what these results meant in terms of human teeth and more particularly in terms of dental caries. It might be thought that the relation of the facts concerning structure and caries could easily have been pursued by further laboratory experiments on animals. But one great obstacle stood in the way of this, namely, that in spite of many attempts it was at the time found impossible to produce dental caries at will under controlled conditions in the animals available. Most of the earlier work on dental caries had therefore to be done on human beings 4,6.

It seemed on a priori grounds likely that if human teeth were badly formed they would be more susceptible to caries, and since this disease is widespread it would logically follow that many teeth would be badly formed. Against this idea was the generally accepted view among dental authorities that although there is widespread caries, human teeth, especially deciduous teeth, in this country are on the whole well formed. Thus an impasse was present and either the hypothesis was incorrect or the accepted views on dental structure were incorrect.  To see whether the structure of teeth was really as good as was stated it was decided first to make a microscopic examination of large numbers of children's teeth. For this purpose, deciduous teeth were collected from various parts of the country and representative of the different classes of society, and after being ground down were classified according to their minute structure. In Fig. 17 photomicrographs of a perfect tooth (a) and a hypoplastic tooth (b) are seen. The following results were obtained :—

Structure of the dentine of British children's deciduous teeth





Grades of dentine structure


Slightly hypoplastic

Moderately hypoplastic

Very hypoplastic

















1st molars






2nd molars






All types






It is clear from the above table that the teeth of British children are, on the whole, very badly formed and that the previously accepted views on this point are wrong. The incisors which are most influenced by antenatal nutrition are better calcified than the later developed molars. It was also found that, on the whole, the structure of the teeth from private sources was better than that of children attending public elementary schools. There seemed, therefore, some basis for the hypothesis that dental structure and caries might be related.

Having determined the structural defects by microscopic examination, this side of the problem was further developed by relating it to the external appearance of teeth (Fig. 18), Thus, the texture of the enamel as indicated by colour and smoothness was carefully noted in a large series of teeth, and then these same teeth were ground down and subjected to microscopic examination. After much experience it became possible to gauge approximately the minute structure by careful examination of the superficial enamel appearance.

Relation of Dental Structure to Caries

A systematic examination was next made, with the idea of seeing the relation between structure (microscopical and superficial) and caries in human teeth 2,5,8. The following tables show some of the results obtained.

Relation of Microscopical Structure to Caries
(All types of teeth taken together.) Deciduous teeth (1500 sections) 

Degrees of caries

Grade of Dentine Structure.

Slightly hypoplastic
Moderately hypoplastic
Very hypoplastic
  % % % %
0 77.9 45.7 8.0 6.0
1 8.1 23.8 14.2 4.9
2 6.5 17.2 33.2 15.2
3 7.5 13.3 44.6 73.9

Note--Some foreign teeth included in this table

Relation of Surface Enamel Structure to Caries
(Ertimated while Deciduous Teeth still in the Mouth)

Degrees of caries

Grade of Dentine Structure.

Slightly hypoplastic
Moderately hypoplastic
Very hypoplastic
  % % % %
0 94.0 74.0 43.0 18.0
1 5.0 16.0 20.0 13.0
2 1.0 9.0 22.0 17.0
3 0.0 1.0 15.0 52.0

From these tables it is obvious that there is a close correlation between structure and caries whether the determination of tooth structure is made by the microscope or by the texture of the enamel as revealed by a superficial examination:  that teeth which are of bad structure tend to be more carious, and those of good structure less carious. This applies to about 90 per cent. of the teeth examined histologically. On the other hand, there is a 10 per cent. disagreement, and this is too large a figure to be regarded as fortuitous. The immediate object of the investiga­tion now was to find the reason for this discrepancy, and back again it came to the laboratory.

The Defence of Teeth

Since the days of John Hunter it has been known that when the enamel and dentine are injured by attrition or caries, teeth do not remain passive but respond to the injury by producing a reaction of the odontoblasts in the dental pulp in an area generally corresponding to the damaged tissue and resulting in a laying down of what is known as secondary dentine. In 1922 M. Mellanby proceeded to investigate this phenomenon under varying nutritional conditions and found that she could control the secondary dentine laid down in the teeth of animals as a reaction to attrition both in quality and quantity, independently of the original structure of the tooth (Fig. 19) 7,8,9. Thus, when a diet of high calci­fying qualities, ie., one rich in vitamin D, calcium and phosphorus was given to the dogs during the period of attrition, the new secondary dentine laid down was abundant and well formed whether the original structure of the teeth was good or bad (Fig. 19 (a)). On the other hand, a diet rich in cereals and poor in vitamin D resulted in the production of secondary dentine either small in amount or poorly calcified, and this happened even if the primary dentine was well formed (Fig. 19 (b) and (c)).

These results showed that teeth, apart from their primary structure, have a second line of defence against injury, one in fact which is dependent on its nutritional supply. Now caries, like attrition, is a harmful stimulus and human teeth attacked either by attrition or caries respond by the production of secondary dentine varying in character (Fig. 20), just as the dog's teeth make a varied response to attrition according to the nature of the diet. Thus it seemed possible that the 10 per cent. exceptions alluded to above might be explained by this second factor of defence. For instance, a badly-formed human tooth might be free from caries because its defence had been increased by good calcifying factors of a nutritional nature. If so, it would produce in response to attrition abundant well-formed secondary dentine.

The result of the animal investigations again brought the subject back to the clinical side in order to see whether badly-formed teeth free from caries contained good secondary dentine and well-formed human teeth which were carious had deficient or badly-formed secondary dentine. This proved to be the case, so that ultimately .the presence of caries in any tooth could be explained on the basis of (a) the original structure of the tooth, or (b) the reaction to injury as disclosed by the condition of the secondary dentine.

It will be obvious that a fact of even greater import could be deduced from these laboratory observations on the condition of secondary dentine for, if true, they foretold the possibility of directly controlling the onset of dental caries in human teeth, independently of their original structure. If, for instance, a child had badly-formed teeth, very susceptible to caries, a diet of high calcifying qualities ought to diminish the susceptibility to the disease. On the basis of this implication M. Mellanby, with the co-operation of C. Lee Pattison, M.B., and J. W. Proud, L.D.S., began in 1921 a series of clinical investigations on children to whom diets of varying calcifying qualities were given, and the rate of development of caries was tabulated. Four of these investigations were made in Sheffield, and were published in 1924, 1926, 1928 and 1933.10 In 1928, also, a large-scale test was begun under "ie auspices of the Medical Research Council at Birmingham,11 with the assistance of Mr A. Deverall, L.D.S., and Miss M. Reynolds. A summary of the findings is given in Table IV.

Synopsis of results of tests made in Sheffield and Birmingham to see whether dental caries in children can be influenced by diet


Percentage of Teeth Carious (Deciduous and Permanent

Percentage increase in “degree” of caries (A.C.F)

First inspection

Final inspection



Olive oil group  





Cod-liver oil group





Vitamin D group






Olive oil group  





Cod-liver oil





Vitamin D group





These results show that the incidence of dental caries can be greatly reduced by feeding children on diets rich in vitamin D. Since, however, the animal experiments had shown that there was another side to the problem of teeth calcification and reaction, namely, that cereals interfere with the processes, it seemed likely that, if a diet could be made not only rich in vitamin D but at the same time be deprived of cereals, the teeth of children eating such a diet might offer still greater resistance to caries. In order to test this possibility, 22 children of average age of 5-4 years were placed on a cereal free diet for 26 weeks (M, Mellanby and C. L. Pattison). The teeth were charted and the number and extent of all carious points noted at the beginning and end of the feeding period, so that any increase in incidence or spread could be determined. Note was also taken of the texture or "hardness" of the carious area at the beginning and end of the feeding period, so as to get some idea of the degree of arrest of the carious process.

The results are given in Table V. (Diet 8), and, with them, corresponding results in other groups of children: (1) where vitamin D only had been added to the diet, and (2) where no extra fat-soluble vitamins had been given, but some oatmeal had been included in the diet.

The Effect of a Diet rich in Vitamin D and devoid of Cereals on the Incidence and Spread of Dental Caries in Children under Six Years of Age.


Diet 8
 Vitamin D

Diet 7.
Rich in Vitamin D.   Contained Cereals

Diet 4
Contained no extra Vitamin D.
Contained oatmeal

Number of children in group    




Average age of teeth per child showing initiation or spread or caries




Average number of teeth per child in which caries showed hardening 




It will be seen from these figures—

(1) that the addition of vitamin D greatly diminished the spread of caries and caused increased arrest of this process (Diet 7).

(2) that the removal of cereals together with the addition of vitamin D virtually suppressed all dental caries and increased the healing process (Diet 8).

Although Diet 8 contained no bread, porridge or other cereals, it included a moderate amount of carbohydrates, for plenty of milk, jam, sugar, potatoes and vegetables were eaten by this group of children. Full details of the constitution of this cereal-free diet can be seen in the publication of M. Mellanby and C, L. Pattison describing this investigation10(d). It is of interest to note that these results are in harmony with those of Boyd and Drain,12 who found that caries in the teeth of children on diabetic diets (devoid of cereal) did not spread or develop.

The hardening of carious areas that takes place in the teeth of children fed on diets of high calcifying value indicates the arrest of the active process and may result in “healing” of the infected area. As might be surmised, this phenomenon is accompanied by a laying down of a thick barrier of well-formed secondary denture. .Illustrations of this healing process can be seen in Figs. 21 (b), (c) and (d). Summing up these results it will be clear that the clinical deductions made on the basis of the animal experiments have been justified, and that it is now known how to diminish the spread of caries and even to stop the active carious process in many affected teeth.

Here, then, we have an example of an investiga­tion in which the laboratory and the clinical work have interacted like battledore and shuttlecock in the following way:—

In the laboratory the dietetic and environmental conditions for the production of perfect and imperfect teeth were found.

On the clinical side, the structure of human teeth was investigated, and was found to be commonly bad, with defects, on the whole, similar to those produced in the experimental animals. It was then found that, generally speaking, well-formed human teeth are less, and badly-formed teeth more, susceptible to caries. This rule was not without exceptions, and thus the investiga­tion came back to the laboratory to find the reason for these exceptions.

The laboratory investigation led to the discovery that teeth, independently of structure, had a power of resistance to harmful stimuli (attrition) which could be largely controlled by diet.

Applied to the clinical side, this explanation was found to cover most of the exceptions in the relation of structure to caries mentioned above. Extended further in the clinical aspect, it was shown that the resistance of children's teeth to caries could be directly controlled in­dependently of original structure by feeding on special diets after full eruption of the teeth. It was, moreover, shown that active carious areas could often be "arrested," with the production of much secondary dentine, by giving the children diets of very high calcifying activity, i.e., rich in vitamin D, calcium and phosphorus. If cereals were also eliminated from the diet, caries was virtually stopped.

The etiology of Periodontal Disease (Pyorrhoea)

Before leaving this subject of the teeth, I should like to refer briefly to another aspect of dental disease, namely, periodontal disease (pyorrhoea). Unfortunately, from the point of view of the present discourse, the work has not yet been extended to pyorrhoea in man, but the experimental results obtained in animals are so clear and, at the same time, so important that they may well be mentioned, especially as they have some bearing on the question of nutrition and infec­tion discussed in Chapter IV.

M. Mellanby has shown that whereas vitamin D is mainly responsible for the calcifying process in teeth and alveolar bone, vitamin A controls the development of the gingival epithelium. If, for instance, young dogs are fed on diets deficient in vitamin A, the stratified epithelium becomes hyperplastic and over­grown. This can be seen clearly in Figs. 22 and 23, where the normal gingival epithelium of a dog fed on a diet rich in vitamin A (a) is compared with the hypertrophied epithelium of a dog on a vitamin-A-deficient diet (b). When epithelium is once laid down in this hypertrophied fashion, it tends to become infected with micro-organisms. This process seems almost inevitable for, after a period of defective feeding, a good diet very rich in vitamin A may not prevent infection. On the other hand, if the epithe­lium in the growing animal is well formed, a defective diet later in life is not so likely to allow the infective process to develop. It is, therefore, most important in the case of the gingival epithelium, just as in that of the teeth themselves, to see that dietary conditions during growth produce perfectly formed tissues.

It is probable that cereals also play a part in inducing the defective formation of the epithelium and the tendency to pyorrhoea, just as they do in the case of dental structure and caries.

The prevention and development of pyorrhoea in older animals brought up under controlled dietary conditions can be seen in the radiographs in Fig. 24 (a) and (b).


1.   Mellanby, M.     (a) Lancet. Lond.. 1918,2, 767; (b) Dent. Rec., 1920, 40, 63.
   Mellanby, M.     Brit. Dent. J. 1923, 44, 1031.
  Mellanby, M.   Sp. Rep. Ser. Med. Res. Coun. Lond.,1929, No. 140.
Mellanby, M.  (a) Brit. Dent. J., 1923, 44, 1; (b)Brit Dent. J. 1927, 48, 737.
Mellanby, M.  Sp. Rep. ser.Med. res. Coun. Lond,1934, No. 191.
  Mellanby, M.  Brit. Dent. J., 1927, 48, 1481.
Mellanby, M.  Proc. R. Soc. Med., 1923, 16, 74-82 (Sect. Odont.).
Mellanby, M.  Brit. Dent. J, 1928, 49, 769.
Mellanby, M.  Sf. Rep. Ser. Med. res. Coun. Lond,1930, No. 153.
10. Mellanby, M., Pattison, C. L., and Proud,W (a) Brit. Med. J. 1924,2, 354; (b) Brit Dent. J 1926, 47, 1045; (c) Brit. Med. J, 1928, 2, 1079; (d) 1932, 1, 507.
11.   Dental Disease Committee Birmingham Int. Rep., Sp. Rep. Ser, Med. Res. Coun. Lond.
12.   Boyd and JD Drain, C.L. J. Amer. Med. Ass., 1928, 9
0, 1867.