throat cancer, through all 120-odd hours of the reading. But that's water under the bridge. I took the (immense) trouble to teach the well-meaning but naive programs improved pronunciations of over 4.000 words, mostly proper names, British spellings, and 18th century archaisms. Having some background in Catholic Church Latin, and two years of (forced) Latin in high school, but no classical scholar, I can only vouch that I did the best I could within my hobbyist limitations. In cases of extremely obscure items such as "Goisvintha"—a Gothic queen; or "Geougen"—a collective noun, like "herd" or "squad", to describe a certain group of Tartaric barbarians, I've settled for what sounded, to my part-time poet's ear, euphonious and consistent with the flow of the line. I've heard tapes of classical scholars rendering academically correct Latin, but, to venture arrogance, neither they nor the Church ever convinced me that anyone except prelates, scholars, and perhaps an affected upper crust ever actually spoke that way. How else explain French? I'd give a lot to bounce through Latium once per century from 600 B.C through the first millennium of our era in order to hear authentic village or "street" Latin.

One notices that at Chapter 39 a different voice takes over through to the end of the work. For this there are two major reasons. Firstly, the product of the original voice required some extensive, and excruciatingly time-consuming digital audio processing to remove crudities of the speech engine, such as a harsh rasp, which even yet breaks through on occasion. Secondly, the production of all six volumes occupied the occasional parts of three or four years, during which time a much improved speech engine became available. Back then I did not envision a public audience for the work, and so was not hobbled by considerations of consistency. The new voice was smoother, endowed with a much more "intelligent" engine, AND DID NOT CRASH EVERY TWENTY MINUTES! For those interested, the softwares are the two final versions of Monologue, created by First Byte Software. I do not believe it is any more available to the market.

There are various types of flaws in this rendering, some correctible, some not—at least not very easily or conveniently. Certain errors persist because I recall they exist, but not where. Now, I don't have the time or inclination to reaudit all 120 hours of the reading in order to flush them out. You are welcome to try, or even to email me, at the address below, their nature and whereabouts. I will collect such reports and maybe someday repair the faults. Better yet, do it yourself (See below)! There are a very, very few errors of garbled audio that arose during the conversion to .mp3. I do not believe they inhibit understanding of the contextual passage. But once again, you tell me. Other errors arise because the programs were not smart enough to recognize and discriminate among some of the tricky ambiguities and homonymic mimeses of English. Sometimes "lives" is pronounced incorrectly relative to context, noun versus verb. Other examples in this category are "minute", "rebel(s)", "lead", "read", and "present". There are occasional renderings of "…part i" -(roman numeral one) as "…part eye". I tried to eliminate that by reformatting the source text, but obviously missed in some cases. The historian Procopius' name is occasionally rendered "Procompanyius" because of an obstinacy about expanding abbreviations I could not completely—Gibbon's favorite term—extirpate. In some few other cases there are obviously erroneous or awkward pronunciations of words, especially of French proper and place names. Frankly, I'm proud the little beasts do as well as they do in the continental tongue. Let me say that although I tried assiduously, and even systematically to eliminate errors (having written a program that scanned the source text for words "new" to the speech software), the product of Gibbon's twenty year labor is in itself an empire of English, in