The alphabetical method did not hold its own without struggle. It prevailed in Robert Stephanus' Latin _Thesaurus_ (1532), the most considerable work of its kind that had been compiled since the invention of printing; but Dolet's Commentaries on the Latin Tongue (1536), are practically a reversion to the arrangement by roots. Henry Stephanus' Greek _Thesaurus_ (1572) and Scapula's well-known abridgement of it (1579) are both radical; and as late as the seventeenth century this method was employed in the first Dictionary of the French Academy, which was designed in 1638 but not published till 1694. That, however, was its last appearance. The preface to the Academy's second Dictionary (1700 and 1718), after comparing the two methods, says: 'The arrangement by roots is the most scientific, and the most instructive to the student; but it is not suited to the impatience of the French people, and so the Academy has felt obliged to abandon it.'[11] The ordinary user of dictionaries to-day would be surprised at being called impatient for expecting the words to be put in alphabetical order. [11] Cf. R.C. Christie, _Etienne Dolet_, ch. xi. In mediaeval times there was one very real obstacle to the use of the alphabetical method, and that was the uncertainty of spelling. Both Papias and Balbi allude to it in their prefaces; but it did not deter them from their enterprise. Even in the days of printing language takes a long time to crystallize down into accepted forms, correct and incorrect. You may see Dutchess with a t at Blenheim, well within the eighteenth century, and forgo has only recently decided to give up its e. In the days of manuscripts men spelt pretty much as they pleased, making very free even with their own names; and uncritical copyists, caring only to reproduce the word, and not troubling about the exact orthography of their original, did nothing to check the ever-growing variety. Such licence was agreeable for the imaginative, but it made despairing work for the compilers of dictionaries. Some of their difficulties may be given as examples. In the early days of minuscule writing, when writing-material was still scarce, to save space it was common to write the letter e with a reversed cedilla beneath it to denote the diphthongs -ae and -oe. In the Middle Ages the cedilla was commonly dropped, leaving the e plain; and so mostly it remained until the sixteenth century revived the diphthong, or at least the two double letters. At all periods down to 1600, some hands are found in which it is impossible to distinguish between c and t; and hence in mediaeval times, and even later, such forms as fatio, loto, pecieris, licterae are not infrequently found for facio, loco, petieris, litterae. An extreme example of the confusion which this variability must have caused is in the case of the fourteenth-century annalist, Nicholas Trivet, whose surname sometimes appears as Cerseth or Chereth.