The Lost Empire Of Atlantis
Gavin Menzies
(Swordfish, London; 2011; pp. xxvi + 374)


The expansive claims of Gavin Menzies about 15th-Century Chinese explorations around the world have been extensively critiqued; the linguistic aspects of his case were reviewed by me in Skeptical Adversaria in 2010. Menzies has now published this book, in which he argues (against all mainstream scholarly opinion) that the fabled island of Atlantis – initially described by the ancient Greek philosopher Plato as part of a moral tale – really existed as an advanced, far-flung civilisation, centred on Minoan Crete.

In support of his case, Menzies draws on the work of both scholars and non-mainstream writers, not always adequately distinguishing between these two groups of sources. For example, his list of sources on pp ix-xiii includes figures such as the hyper-diffusionist Gunnar Thompson; he also fails to make explicit the relative even-handedness of authors such as John Sorensen, who does not endorse all of the many claims which he surveys.

Again, I criticise here the linguistic aspects of Menzies’ theory; it is possible that the other aspects of his proposal (see below) are more persuasive. Menzies’ main source on Cretan languages and scripts is Minas Tsikritsis (see http://www.anistor.gr/english/enback/v014.htm).

The three most important early Cretan scripts (found on clay tablets) are: ‘Linear A’ (the number of distinct symbols suggests that this script is ‘syllabic’ – one symbol per syllable – but most mainstream authors regard it as so far undeciphered; the ‘maverick’ professional linguist Cyrus Gordon analysed the texts as being in a West Semitic language – fairly similar to Phoenician and Hebrew – but although this is not historically implausible his ‘decipherment’ has not been generally accepted); the later, superficially similar ‘Linear B’ (deciphered as a syllabic representation of early Greek by Michael Ventris and John Chadwick in 1952; some historians were initially surprised by this identification, but almost all scholars now accept the decipherment); and the script used in the mysterious text on the Phaistos Disk (discussed by me in Skeptical Adversaria in 2008), which displays around 45 distinct symbols and is thus probably in a syllabic script but is almost certainly too short (around 240 symbol-tokens) to be persuasively deciphered.

Menzies’ main discussion of Tsikritsis’ ideas is on pp. 314-321. Tsikritsis (himself from Crete) believes that he has deciphered Linear A as an early form of Greek. Others, notably Steven Fischer, have advanced similar claims regarding the Phaistos Disk, and indeed Tsikritsis himself regards fifteen of the symbols on the Disk as shared with Linear B and ‘deciphers’ part of the Disk text too as Greek. But these ‘decipherments’ of both bodies of material require otherwise unknown (and in places unlikely) varieties of Greek (unlike the Ventris-Chadwick decipherment of Linear B, which actually confirmed earlier philological predictions as to the pronunciation of Greek in times before the alphabet was adopted). Furthermore, Linear A in particular would probably have already been reliably deciphered on the basis of Linear B if it did indeed represent Greek; this has in fact been attempted a number of times, but without demonstrable success. In fact, many writers with varying degrees of expertise have offered speculative decipherments of Linear A (as of the Disk) as representing a plethora of languages (for instance, Leonard Palmer, discussed by Menzies on p. 64, believed that he had discerned links with the Anatolian language Luvian); and anyone proposing a new interpretation needs to argue that his own decipherment should be preferred to these earlier efforts – and to the mainstream ‘null hypothesis’ that no decipherment currently recommends itself. Tsikritsis has not been able to persuade mainstream scholars that he has achieved this.

Supported by sources such as Hans Peter Duerr (see p. 97), Tsikritsis also proclaims that various bodies of symbols found in various locations spread across Europe, the Near East, India, etc. represent Linear A, and thus indicate (along with his readings of the Cretan texts) that the users of the script operated far beyond Crete and the Aegean. However, the evidence for these identifications appears inadequate. The parallelisms are not patently systematic, and indeed the bodies of non-Cretan data are typically too small for systematicity to be manifested; for instance, Duerr cites a single isolated symbol (from northern Germany).

Tsikritsis also holds (citing Gordon in support; see p. 260) that the Linear A texts contain much mathematical symbolisation; and he extrapolates from his ‘findings’ to argue that the Minoans started the Olympian Games, invented mechanical ‘computers’ and in general constituted a ‘lost’ ancestral civilisation of vast sophistication, which can be identified with Atlantis.

Menzies himself makes some dubious theoretical/methodological claims. For example, he states (p. 316) that 56 distinct symbols are required if a language is to be translated ‘with absolute statistical certainty’ (and that the Phaistos Disk script thus displays marginally too few symbols for this). I do not know where he obtained this figure of 56. Any such calculation would depend on the total predicted number of the symbols in the entire writing system, and this in turn depends on the script-type involved: logographic (thousands of symbols, as in Chinese script), syllabic (typically between 40 and a few hundred), or alphabetic (between 10 and 150). More importantly, the likelihood of arriving at an authoritative decipherment of a particular text or body of texts crucially involves (as noted above) the volume of material available (that is, the combined length of the relevant texts). And there are other factors too, notably the amount of supporting cultural information available, the presence or absence of potentially recognisable proper names or of other words likely to be shared with known languages, the historical plausibility of each proposed identification, etc., etc. Menzies has grossly oversimplified this issue.

If Menzies’ theories are to be accepted, it must be on the basis of his non-linguistic evidence; he especially emphasises genetic data. I will not attempt to evaluate this evidence here.