Borrowed or Inherited? Mistake Gives a Clue.

The modern speakers of Misima, an Austronesian language right down on the New Guinea Bird's Tail use the following words in counting nowadays:
6 esiwa, 7 ewon, 8 epit, 9 ewata.

The same words were recorded a century ago, as
6 siwa, 7 on, 8 pit, 9 ata,
but only in counting 10s, ie in 60, 70, 80, 90 while the 'lower number' words in Misima for 6-9 were the usual hand-1, hand-2, etc., common in most An groups in that area.

I glanced at them (they're very familiar Austronesian number words),and thought, well that blows my theory that number words were conceived a long time later than the reconstructed proto-Austronesian
*enem, *pitu, *walu, *siwa (in that order).

Perhaps the proto-Austronesians did, after all, have a decimal system, and the more primitive systems in New Guinea, etc, really are `retrograde' systems brought on by Papuan influence, which is the conventional linguistical view.

But the Misima are the only Austronesian group in that area that has these words, so I looked again a bit harder.

And suddenly realised they were using the right words OK, but in the wrong places.
The Misima use 9 for 6, 6 for 7, 7 for 8, and 8 for 9.
Nobody else, anywhere, does that.

It almost proves that these number words were borrowed, not inherited.

And it also almost proves that someone taught them how to make number words for the higher decades, but not how to use them correctly, and they still don't use them properly.

Link to Austronesian Numbers Worksheet

I' ve posted the current version of my worksheet on the web at:
http://coconutstudio.com/Austro%20Nos%20Mar2008%2011-3.xls

So far, I've listed some 1600 number systems in both Austronesian and Papuan languages, and analysed them as best I can, with a code system that reduces a mass of information to a manageable size.

Examples:
Maisin 6 = faketi tarosi taure sese which means 'hand over 1' is coded 5\1 because faketi tarosi is 5, the \ stands for a regular 'connector' and sese means 1.

Arifama-Miniafia, another An language close by, has 6 = umat roun ta'imon where 5 = umat roun , so that's coded 51 because there is no 'connector'.

Another dialect of Arifama has 6= uma ti reban taimo nomon, 5 = uma ti morob, and 5 isn't repeated exactly in 6, while uma means hand, so this is coded H\1 (or should have been, but I made a typing error here) .

The coding has made it much easier to visualise connections between number-types in various language sub-groups (and whether they match up or not) and their distribution over larger areas.

This informaion (when I've worked out how to use Photo-Shop) will be transferred to geographical maps, making the picture a whole lot clearer.