Anyone familiar with the phrase 'caveat emptor'?

Tore Janson's book purports to be "the first ever popular history of Latin". For "popular", read "dumbed down". The misconceived claims that mar it include the comforting assurance that "anyone who has a large vocabulary in English" already "knows quite a lot of Latin words". This is rather like arguing that, although you may never have heard of al-Khwarizmi, you already know quite a lot about Arabic mathematics by virtue of being able to count on a decimal scale in your own language.

On the first page, Janson says "it is easy to guess" that the Latin word "femina" means "woman". An easy guess if you already assume that English words such as "feminine" and "female" are somehow derived from Latin. The trouble with Janson's simplistic assumption is that your easy guess could easily be quite wrong: "femina" might mean not "woman" but "wife", "girl", "matron" or even "mistress". Such a sloppy approach to etymology does no service to students seriously interested in linguistics.

For whom, then, is the book intended? The target readership, it emerges, apparently includes those naive enough to believe that "anyone who has mastered Latin would be able to recognise all the words in French" (wrong: no marks) or that Latin is "the world's most successful language" (right: ten out of ten). Janson somehow overlooks the fact that, by his own criteria, the most successful language in the world so far must be English by a long chalk.

All he shows (without difficulty) is that Latin long ago supplied many lexical borrowings into other languages. It goes without saying that countless European writers were brought up on acquaintance with the Latin classics. But that is a different proposition altogether.

The translators and their publisher are evidently trying to present Janson's text as an elementary course book for the British/ American undergraduate market, geared towards those whose first-hand acquaintance with Latin verges on zero. That suspicion is borne out by the inclusion of a section on Harry Potter. There, we learn, unsurprisingly, that J.K.

Rowling's Latin is bogus. It "turns out to be mostly English in disguise".

This, presumably, is a combined tribute to two successful languages rolled into one.

A conspicuous lacuna in a "natural history" of any language is the absence of comparison between its earlier and later forms, spoken or written. There is a mini-grammar and a word list for Classical Latin. But Janson gives us no idea of what distinguishes the Latin of the Praeneste fibula from the Latin of Descartes (the former is never mentioned and the latter, it seems, appears only because some contemporaries called him Cartesius). Nor do we learn much about the late survival of Latin in Poland or its academic appearance in Western doctoral dissertations of the 19th and 20th centuries.

No less patchy is the account of the relationship between political history and the fortunes of Latin. "Augustus ruled for a long time and by and large he ruled well." Did anyone ask the slaves? Janson tells us authoritatively that "no one knows why the Roman Empire collapsed". That must be a relief to those who used to sit entrance exams for Oxford and Cambridge universities. Nor, it seems, does anyone know why Latin survived in some parts of the Roman Empire but not in others. Here, a few maps would have been useful, but none is supplied.

Then there is the vexed question of Latin and Christianity. The religion "became stronger at about the same rate as the Roman Empire grew weaker".

But is this true? If so, where is the evidence? The conversion of Constantine gets one brief sentence, and Justinian fails to make an appearance. Latin, we are told, was intimately bound up with the perpetuation of magic. There were Latin treatises on alchemy. Cato recommended chanting gibberish spells in Latin. And "Latin literature is full of references to wizards and witches". We are invited to draw our own conclusions.

Janson plunges even more recklessly into troubled waters in the long-running debate about the transition from Latin to Romance. He fiddles with the question of what people "thought" they were speaking. Was it Latin or something else? That, he supposes, can be reduced to the names contemporary writers gave to their own tongue ("lingua romana", "lingua romana rustica" and so on). Then comes his astonishing conclusion: "As far as Latin is concerned, it is not very important what people thought they were speaking."

If, as a linguist, you can believe that, then you can believe anything. Or at least you can set yourself up as a "popular" historian of Latin, for whom it matters little what that language meant to its users or why. The book concludes with a glossary of well-known Latin tags, irritatingly omitting their exact sources. The most relevant, in this instance, is "caveat emptor".

Roy Harris is emeritus professor of general linguistics, Oxford University.