As to the particular process he employed, we have, as Mr. Blades points out, only negative evidence on which to rely. The frequent unevenness and irregularity of his lines, as well as the variations of the letters themselves, lead to the conclusion that the method employed was a rude one, inferior not only to that now in use, but even to that adopted by the advanced German School of Typography of his own day. Rude, however, as his method may have been, we are not disposed to allow that Caxton could have produced the types he did without the use of a matrix and an adjustable mould. Despite his rough workmanship, his types are as superior to those of the _Speculum_ and _Donatus_ as they are inferior to those of the _Mentz Bible_ and the _Catholicon_; and we consider it out of the question that works like the _Dictes_, or the _Polychronicon_, or the _Fifteen O’s_, could have been produced from types cast by a clay or sand process, which we have elsewhere described as possibly employed in the most primitive practice of the art. It is more probable that both Colard Mansion and Caxton, possessing the principle of the punch, matrix and adjustable mould, but ill-furnished with the mechanical appliances for putting that principle into practice, made use of rough and perishable materials in all three branches of the manufacture. Some such rough appliances we have already suggested in our introductory chapter. . His {86} punches, as Mr. Blades has pointed out, were, in the case of at least two of his founts, touched-up types of a fount previously in use. A matrix formed from such a punch, either in soft lead or plaster, could not be anything but rough and fragile; and such a matrix, when justified and applied to a mould of which the adjustable parts may have lacked mathematical finish and accuracy, could scarcely be expected to produce types of faultless precision. As we have freely admitted, it is impossible on this subject to go beyond the regions of speculation, but we decidedly incline to the opinion that the irregularities and defects of Caxton’s types may be accounted for in the way here suggested, rather than by the assumption that he made use of a method of casting differing wholly in principle from that which was presently to become the universal practice. We shall now briefly follow Mr. Blades’ chronological summary of Caxton’s six types, with a view to point out such particulars respecting them as may have special bearing on the object of this work.