2. Some errors would, of course, arise from the mere inattention, carelessness, or the ignorance of transcribers; and yet fewer, probably, than may at first be imagined; for besides that those who spent their lives in this occupation would generally acquire a high degree of technical accuracy of eye, ear, and hand, and that correctness and legibility must have been the qualities upon which, principally, the marketable value of books depended; it is known that in the monasteries, from whence the greater part of all existing manuscripts proceeded, there were persons, qualified by their superior learning for the task, whose office it was to revise every book that issued from the Scriptorium. Errors of inadvertency must, nevertheless, have occurred. If the author to be transcribed was read by one person, while several wrote from his voice, the process would be open, not only to the mistakes of the reader’s eye, and to those of the writer’s hand; but especially to those of the writer’s ear; for words, similar in sound, might often be substituted, one for the other. Instances of this sort are of frequent occurrence, and the knowledge of the probable cause often serves to suggest the proper correction. If the writer read for himself, he would be liable to mistake letters of similar shape--to mistake the _sense_ by a wrong division of words in his manner of reading, in consequence of which he might involuntarily accommodate the orthography or the syntax to the supposed sense. The frequent use of contractions in writing was a very common source of errors; for many of these abbreviations were extremely complicated, obscure, and ambiguous, so that an unskilful copyist was very likely to mistake one word for another. No parts of ancient books have suffered so much from errors of inadvertency as those which relate to _numbers_; for as one numeral letter was easily mistaken for another, and as neither the sense of the passage, nor the rules of orthography, nor of syntax, suggested the genuine reading, when once an error had arisen, it would most often be perpetuated, without remedy. It is, therefore, almost always unsafe to rest the stress of an argument upon any statement of numbers in ancient writers, unless some correlative computation confirms the reading of the text. Hence nothing can be more frivolous or unfair than to raise an objection against the veracity or accuracy of an historian, upon some apparent incompatibility in his statement of numbers. Difficulties of this sort it is much better to attribute, at once, to a corruption of the text, than to discuss them with ill-spent assiduity.