As is well known, there has been a good deal of discussion about the unitary or binary nature of depression. Mapother (1926) and Lewis (1934) made a strong case for the unitary view on clinical grounds, Curran (1937) concurring. Kendell (1968) has reviewed the history of this argument; it is marred by confusion which has persisted through recent attempts to use statistical techniques of factor analysis and discriminant function analysis in an effort to find a more objective and empirical solution. It is the purpose of this brief note to draw attention to this confusion, to show how it has affected arguments of both adherents and opponents of the binary position, and to argue that the data are in fact in sufficient agreement to make possible a valid answer to both problems. It is suggested that the apparent disagreement between workers such as those of the Newcastle group (Kiloh and Garside, 1963; Carney, Roth and Garside, 1965) and the London (Maudsley) group (Kendell, 1968) is in fact quite irrelevant and is based on a misunderstanding of the statistical properties of factors, a misunderstanding apparently introduced in one of the first studies of this kind to be concerned with the problem of the classification of depressive illness, that by Hamilton and White (1959).

The relationship between anxiety state and depressive illness has been debated for many years. On clinical grounds, Mapother (1926) and Lewis (1966) expressed the view that the various forms of anxiety and depressive states merge imperceptibly into one another, while Garmany (1956, 1958) and Stenback (1963) considered that they were fundamentally different disorders. Gurney, Roth and Garside (1970) showed that the two conditions differ significantly in respect of a large number of biographical, personality and clinical indices. They also demonstrated that the disorders could be separated by means of a discriminant function analysis and that the distribution of the patients' scores was clearly bimodal. In a four year follow-up study of the same group of patients, Kerr, Gurney, Schapira and Roth (1970) showed that depressive illnesses carry a significantly better prognosis. Moreover, physiological evidence based on forearm blood-flow measurements has also given tentative support to the differentiation of the disorders (Kelly and Walter, 1969).

Proof that two clinical syndromes are distinct entities depends on a demonstration that patients with features of both syndromes are less common than those with features only of the one or the other. The intermediate forms, the greys, must be shown to be less numerous than the blacks and the whites, which means in graphical terms that a bimodal distribution of scores must be demonstrated on some chosen dimension. The optimal dimension for this purpose is Fisher's discriminant function (Fisher, 1936; Rao, 1948), and it may be useful before proceeding further to describe how this is derived. From the universe of all patients with syndrome X (psychotic depression) or syndrome Y (neurotic depression) every patient is assigned to one or other category, and the two populations are then rated on a series of N items which comprise the recognized discriminators between the two syndromes. From these data the analysis produces a set of weights for the N items which maximizes the ratio of between group to within group variance. The effect of this is that, when a single weighted score is calculated for each patient by combining the weights of the relevant items, the overlap between the scores of members of the two populations, X and Y, is reduced to a minimum. If the distribution of the scores of X and Y combined is bimodal the validity of the distinction between the two is confirmed.

The distinction between schizophrenic and affective illnesses has been one of the cornerstones of psychiatric classification ever since Kraepelin introduced the twin concepts of dementia praecox and manic depressive psychosis at the turn of the century. It has also long been recognized that some patients have both schizophrenic and affective symptoms, and various interpretations have been placed on these mixed states. To some continental psychiatrists they constitute a third group of psychoses distinct from both schizophrenia and manic-depressive psychosis—the degeneration psychoses of Kleist or the cycloid psychoses of Leonhard. By others they are regarded as genuine mixed states, with the implication that elements of both schizophrenia and manic depressive illnesses are contributing, perhaps because the genetic or constitutional endowment is mixed, perhaps because two alternative defence mechanisms are being utilized simultaneously. Often, however, mixed symptomatology is simply ignored, either by discounting the schizophrenic symptoms and focusing attention on the mood change, or, as most American psychiatrists do, by glossing over the affective symptoms and regarding the illness as a form of schizophrenia differing in no significant respect from other schizophrenias.

Primary affective disorder is defined as an illness which is characterized by depressions or manias or both. This illness is seen in an individual who has no history of any preexisting psychiatric illness except for uncomplicated episodes of depression or mania. Clinically there are two types of primary affective disorder. The first is manic depressive disease (bipolar psychosis) which may be considered as a primary affective disorder in which mania is seen; usually, but not always, depression is also seen in this kind of patient. The other type may be called depressive disease; a synonym that might be used is unipolar psychosis; here only depressive episodes are seen (one or more episodes). In addition to the difference in the clinical picture, considerable genetic data exist which indicate that the two illnesses are quite distinct from each other.

Human blood plasma normally contains 500–1,000 mg./100 ml. of lipid material kept in solution by complex formation in various proportions with protein to form lipoproteins of definite size and constitution. Apart from the chylomicrons at least four β-lipoproteins and two α-lipoproteins are recognized, differing from each other in density. The α-lipoproteins correspond to the α-globulins and the β-lipoproteins to the β-globulins as determined by paper electrophoresis. The various lipoproteins have been shown to possess a constant and characteristic pattern both in protein: lipid ratio and in the relative proportions of phospholipid, sterol, sterol esters and glycerides (Jones et al, 1951; Bragdon et al., 1956). The proportions in which the fatty acids are present have also been shown to be constant (Lindgren et al., 1959).

Two reports have shown that a substantial proportion of patients suffering from various psychiatric syndromes have low serum folate concentrations (Hunter, Jones Jones and Matthews, 1967; Carney, 1967). The present investigation was concerned to examine further the prevalence of low serum folate in a series of depressive patients on admission to hospital. As folate deficiency contributes to the development of mental symptoms in epileptic patients (Reynolds, 1968), we also examined the relationship of low serum folate concentrations to the severity of the depressive illness, certain personality traits, and the response to anti-depressant treatment. The relationship of low serum folate to variables which could influence it such as diet and drugs was also examined. In a proportion of these patients serum folate measurements were repeated just before the patient's discharge. Serum B12 concentrations were also estimated.

The presence of a birth order factor in the aetiology of schizophrenia is indicated by evidence of two kinds. The evidence, however, is apparently conflicting.

Though there have been many investigations concerned with sibling order and mental illness there have been relatively few concerned with sibship size. The abundance of sibling order studies is probably due to the widely accepted belief that in such studies comparison with general population data is not necessary. Expected numbers within each birth rank are estimated by the Greenwood-Yule method, proposed in 1914, by which it is assumed that for any given size of sibship siblings should be evenly distributed among birth ranks. In the present population, in which family size has in recent years been decreasing, this assumption is clearly invalid, for the younger members of the larger families of earlier years are more likely to be included in a contemporary sample, the older members by now having died. Other factors influencing sibling position have been more fully discussed by Price and Hare (1969). Thus comparison with general population figures is necessary in sibling order studies, and a report on such a comparison will be published separately by the author.

A repertory grid is obtained when a person compares and rates a set of ‘elements' in terms of an appropriate set of ‘constructs', using a consistent procedure throughout (Slater, 1964). The elements may be the names of real or imaginary people, qualities, or activities, in fact any group of items to all of which each construct is logically applicable.

Repertory grids can be applied to individuals in many different ways (Bannister and Mair, 1968). A grid is obtained whenever an observer rates or compares a set of objects, the elements, in terms of a set of logically applicable notions, or constructs, using a consistent scoring procedure throughout (Slater, 1964).

We have previously reported the use of repertory grid testing to extend the understanding of psychopathology in neurotic patients and to test predictions about the effects of therapy (Ryle, 1967; Ryle and Lunghi, 1969). In these tests the elements were people known to the subject, and the constructs were elicited from him. Repertory grid testing can supplement the data obtained from clinical interviews, notably in the following ways:

We have previously suggested that a number of psychoanalytic concepts have undergone changes of meaning when extended to wider settings both within psychoanalysis and outside it (Sandler, Dare, and Holder, 1970a, b; Sandler, Holder and Dare, 1970a, b, c). This has led to situations in which the same term can have a variety of meanings attached to it, with consequent confusion and loss of precision. A clear example of this is the concept of transference, which is now given a wide range of meanings, including that of being simply a synonym for ‘relationship'—a usage substantially different from its original meaning (Sandler, Dare and Holder, 1970b).