voluntary turning about, means anything from shaking hands with a revivalist, as a sign of religious desire, to the profoundest reversal of emotions and of likes and dislikes. And not only must terms be interpreted by the investigator, but whole situations must be reconstructed by him from given fragments. He must allow for the respondent's probable ignorance of certain factors; for bias produced by training; for the influence of conditions probably present but not proved to be so; for individual peculiarities gathered by the investigator from the tone of the whole response. Such interpretation may be skilfully done, and it may be as worthy of confidence as similar interpretations made by historians, but it requires the greatest caution and balance, and in the end its results, like history, lack exactness.1

On the other hand, it would be hasty to assume that question circulars are scientifically useless. A large part of the data in several scientific fields is, in fact, gathered by question lists called report blanks. Further, there is no absolute separation between questions asked of a subject in a laboratory and questions asked of subjects under other conditions. The main difference is in the degree of our knowledge of the stimuli actually present. In either case we accept more or less of the subject's observation of himself. It should be remembered also that the question circular need not, and should not, be

Starbuck has used his returns with caution and in an objective spirit. He has avoided the worst pitfalls into which question-list researches have fallen. Yet his numerical tabulation of emotions, motives, and the like shows nothing more than general drifts present in unknown proportions.

isolated from other instruments of research. Now and then details can be run down by personal interviews or experiment.1

Question-circular returns are effective in establishing at least three types of generalization: (1) External situations in which some broadly recognized psychical event takes place. Such facts as age, date, and social environment can often be ascertained with approximate accuracy. Largely through this method one great generalization of this type has been established, namely, that there is a causal connection between religious conversion of the emotional type and the physiological change during adolescence. (2) The existence of contrasting types within a specified field. This certainly is a valuable psychological service. In the matter of prayer, for example, and in general in the individual realization of God in a high form of religion like the Christian, there are striking differences of type. Self-assertion characterizes one individual, self-abnegation another. Some persons have a mystical-emotional realization of God, while others find him through abundant, free activity. Further, this method is important because it definitely establishes the existence of contrary reactions in groups, such as religious denominations, that may seem to be homogeneous. (3) The existence of a tendency or drift within a group may be ascertained, even though the extent or depth of the drift may remain unknown. Thus, a question that I once asked concerning the nature of their call to the ministry elicited from a significant number of ministers in a denomination that cultivates emotional realizations the surprising fact that their call I See Coe, Spiritual Life, chap. iii.

contained little of this element, but much of prosaic, common-sense procedure.1

Data for the psychology of religion are gathered, in the second place, by scrutiny of literary or other records of religious life." Here fall biographies and autobiographies; scattered passages in general literature; sacred literatures, which include (often in mixture) history and biography, myth, hymn, prayer, ritual, and theological or metaphysical thought structures;3 finally, inscriptions, pictures, statues, temple architecture, and the like. Such sources for what men tell about their religion have one advantage over the question list, namely, that the investigator certainly does not influence the original records. They have been produced in the ordinary course of life or of religion, and often the absence of psychologizing makes them psychologically

I

To judge the feasibility of a given question-circular enterprise, the following questions will be found useful: (1) How much of it concerns matters in which non-expert testimony is likely to be adequate? ¦ (2) Does it formulate answers, or does it merely suggest situations, leaving the respondent to formulate his response? (3) Does it offer alternative answers or classes into which the respondent is to fit himself? If so, are the alternatives exhaustive and mutually exclusive? (4) Is the language such that the respondent will get the points intended by the questioner? (5) What is the point of the whole list? If trustworthy responses were made in adequate number, how would our knowledge be advanced? Is this knowledge accessible in any more direct or trustworthy fashion?

2 James's The Varieties of Religious Experience is based largely upon autobiographic records. Another specimen is Royce's study of Bunyan in Studies in Good and Evil. A more extensive research based on this method is Delacroix's Etudes d'histoire et de psychologie du mysticisme.

3 Many writers have dipped into the records of the ethnic faiths, but Stratton in his Psychology of the Religious Life bases thereon nearly the whole of a psychological theory of religion.

An example is Harrison's Themis (Cambridge, 1912).

the more valuable. Nevertheless, there are difficulties that must not be ignored.

1. The biographical and autobiographical material that is available is triply selected. First, the individuals portrayed are selected by biographers and autobiographers upon principles that, even where they can be ascertained, have little relation to the psychologist's interest. Secondly, the material that is recorded in each case is also selected from a larger mass, and again from motives that produce no psychological classification. Thirdly, the psychologist selects the writings that he will analyze. James has been criticized, for example, for selecting too large a proportion of extreme and even morbid cases. In short, this material is valuable chiefly because, like question-list returns, it establishes the existence, in unknown proportions, of certain types of religious experience.1

2. Sacred literatures offer the extraordinary advantage of being religiously motived, so that they present to us, as it were, religion itself at some stage of its ongoing. Nevertheless, even here we find comparatively little that is naïvely religious. Men make a written record, for the most part, because they have become reflective and therefore selectors. Much is motived by desire to explain, reconcile, or systematize, as in myth and theology. Some of the records are made in the interest of a cause, an idea, or a party, and consequently the writer maintains silence concerning shady aspects of his own religion, but gives prominence to such aspects of his

1 A. R. Burr, in Religious Confessions and Confessants, is so unconscious of these difficulties as to suppose that a "definitive" collection of data on religious experience could be made from autobiographic confessions.

opponent's religion. Ritual formulae, in turn, commonly give words to be said rather than things to be done, whereas these latter are essential to most rites. Prayers and hymns, on the other hand, sometimes let us see far into the religious mind, but even here the meaning is likely to depend upon historical conditions that are not specified. Indeed, sacred literatures as a whole arise and grow as parts of a historical movement which they only partly reflect. Different historical and literary strata, for example, are often present in an Old Testament story. As a consequence of all this, we can rarely make sure of our psychological data by merely reading a piece of sacred literature. "Blessed are ye poor," says Jesus according to Luke; "Blessed are the poor in spirit,' according to Matthew. What, then, was Jesus' actual point of view? It is scarcely necessary to add that many of these difficulties attach themselves to the study of inscriptions and of religious art.

In the third place, data are found by anthropological research, chiefly among the lowest peoples, but in part also among the less sophisticated ways of culture-races. The amount of material now in print that bears upon the religion of savages may properly be called immense. Much of it has the peculiar value for psychology that it enables us to get near to the beginnings of religion. In the matter of origins, definitely ascertained facts are now taking the place of speculation or of inferences drawn from our own highly developed processes. We are on the way toward knowledge of the evolution of religion that will be comparable to our knowledge of the evolution of man's physical organism. That the road is not a straight and broad one, however, may be