Secondly, as I have already said, Roman evidence is peculiarly difficult to handle, except in so far as it deals with the simple facts of worship; when we use it for traditions, myths, ideas about the nature of divine beings, we need a training not only in the use of evidence in general, but in the use of Roman evidence in particular. Anthropologists, as a rule, have not been through such a training, and they are apt to handle the evidence of Roman writers with a light heart and rather a rough hand. The result is that bits of evidence are put together, each needing conscientious criticism, to support hypotheses often of the flimsiest kind, which again are used to support further hypotheses, and so on, until the sober inquirer begins to feel his brain reeling and his footing giving way beneath him. I shall have occasion to notice one or two examples of this uncritical use of evidence later on, and will say no more of it now. No one can feel more grateful than I do to the many leading anthropologists who have touched in one way or another on Roman evidence; but for myself I try never to forget the words of Columella, with which a great German scholar began one of his most difficult investigations: "In universa vita pretiosissimum est intellegere quemque nescire se quod nesciat."[22] NOTES TO LECTURE I [1] Mommsen, _Hist. of Rome_ (_E.T._), vol. ii. p. 433. [2] Cumont, _Les Religions orientales dans le paganisme romain_, p. 36. Cp. Dill, _Roman Society in the Last Century of the Western Empire_, p. 63. Gwatkin, _The Knowledge of God_, vol. ii. p. 133. [3] See some valuable remarks in Lord Cromer's _Modern Egypt_, vol. ii. p. 135. [4] Since this lecture was written this scholar has passed away, to the great grief of his many friends; and I refrain from mentioning his name.