1. In judging of the external evidence for any ancient writing, it is always important to observe not only the details of the evidence itself (date, genuineness, authority, freedom from ambiguity, the precise point attested), but also the extent of the area from which it is drawn and the proportion which it bears to the extant literature of the period which it covers. The first step should be an attempt to realize by an effort of the imagination the proportion between (1) the whole of the extant evidence, (2) the amount of the material that yields this evidence, (3) the amount of the material, once extant but now no longer extant, which might have contributed evidence if we had it. In other words, what we have to consider is not only the actual, positive evidence available, but the distribution of this evidence and its relation to the real lie of the facts—no longer accessible to us but as they may be imaginatively reconstructed. 2. In particular, when use is made of the argument from silence, the first question to be asked is, What is silent? It may well be that the literature supposed to be silent is so small that no inference of any value can be drawn from it. 3. In any further use of the argument from silence full allowance should be made for common human infirmity in the persons who are silent—for oversight forgetfulness, limited range of thought. It is always desirable that the application of the argument from silence should be checked by comparison with verifiable examples from actual experience, whether that experience is derived from ancient life or from modern. 4. The presumption is that plain statements of fact may be trusted, unless there is a distinct and solid reason to the contrary. Even where there is a considerable interval of time between the fact and the statement, it may be presumed that the writer who makes the statement had connecting links of testimony to which he had access and we have not. In any case it is worth while to ask ourselves whether it is not probable that such connecting links existed. 5. In such plain statements the presumption further is that the writer meant what he says, or appears to say. Not until this apparent sense has proved wholly unworkable is it right to tamper with his express language, whether by emendation of the text or putting upon his words a sense that is not obvious and natural. 6. The imputation of conscious deception or fraud is to be strongly deprecated, except with writers of ascertained bad character, and even then the imputation should not be made without substantial reason. 7. All imputations of motive, and especially of sinister motive, should be carefully weighed, and it should in particular be considered whether the supposed motive is one that was likely to be in operation under the historical conditions of the time and circumstances of the writer affected.