Guns, Germs & Steel

supporting humans. Astonishingly, it also swept westward across the Indian Ocean to the east coast of Africa, resulting in the colonization of the island of Madagascar.

SPEEDBOATTO POLYNESIA • 341

Figure 17.2. The paths of the Austronesian expansion, with approximate dates when each region was reached. 4a — Borneo, 4b = Celebes,4c = Timor (around 2500 b.c.). 5a – Halmahera (around 1600 b.c.).5b = Java, 5c — Sumatra (around 2000 b.c.). 6a = Bismarck Archipelago (around 1600 b.c.). 6b = Malay Peninsula, 6c = Vietnam (around 1000 b.c.). 7 = Solomon Archipelago (around 1600 b.c.). 8 – Santa Cruz, 9c = Tonga, 9d = New Caledonia (around 1200 b.c.).lOb= Society Islands, lOc = Cook Islands, 11 a = Tuamotu Archipelago(around A.D. 1).
At least until the expansion reached coastal New Guinea, travel between islands was probably by double-outrigger sailing canoes, which are still widespread throughout Indonesia today. That boat design represented a major advance over the simple dugout canoes prevalent among traditional peoples living on inland waterways throughout the world. A dugout canoe is just what its name implies: a solid tree trunk 'dug out' (that is, hollowed out), and its ends shaped, by an adze. Since the canoe is as round- bottomed as the trunk from which it was carved, the least imbalance m weight distribution tips the canoe toward the overweighted side.

3 4 2 'GUNS,GERMS, AND STEEL
Whenever I've been paddled in dugouts up New Guinea rivers by New Guineans, I have spent much of the trip in terror: it seemed that every slight movement of mine risked capsizing the canoe and spilling out me and my binoculars to commune with crocodiles. New Guineans manage to look secure while paddling dugouts on calm lakes and rivers, but not even New Guineans can use a dugout in seas with modest waves. Hence some stabilizing device must have been essential not only for the Austrone-sian expansion through Indonesia but even for the initial colonization of Taiwan.
The solution was to lash two smaller logs ('outriggers') parallel to the hull and several feet from it, one on each side, connected to the hull by poles lashed perpendicular to the hull and outriggers. Whenever the hull starts to tip toward one side, the buoyancy of the outrigger on that side prevents the outrigger from being pushed under the water and hence makes it virtually impossible to capsize the vessel. The invention of the double-outrigger sailing canoe may have been the technological breakthrough that triggered the Austronesian expansion from the Chinese mainland.
Iwo striking coincidences between archaeological and linguistic evidence support the inference that the people bringing a Neolithic culture to Taiwan, the Philippines, and Indonesia thousands of years ago spoke Austronesian languages and were ancestral to the Austronesian speakers still inhabiting those islands today. First, both types of evidence point unequivocally to the colonization of Taiwan as the first stage of the expansion from the South China coast, and to the colonization of the Philippines and Indonesia from Taiwan as the next stage. If the expansion had proceeded from tropical Southeast Asia's Malay Peninsula to the nearest Indonesian island of Sumatra, then to other Indonesian islands, and finally to the Philippines and Taiwan, we would find the deepest divisions (reflecting the greatest time depth) of the Austronesian language family among the modern languages of the Malay Peninsula and Sumatra, and the languages of Taiwan and the Philippines would have differentiated only recently within a single subfamily. Instead, the deepest divisions are in Taiwan, and the languages of the Malay Peninsula and Sumatra fall together in the same sub-sub-subfamily: a recent branch of the Western Malayo-Polyne-

SPEEDBOATTO POLYNESIA • 343
sian sub-subfamily, which is in turn a fairly recent branch of the Malayo-Polynesian subfamily. Those details of linguistic relationships agree perfectly with the archaeological evidence that the colonization of the Malay Peninsula was recent, and followed rather than preceded the colonization of Taiwan, the Philippines, and Indonesia.
The other coincidence between archaeological and linguistic evidence concerns the cultural baggage that ancient Austronesians used. Archaeology provides us with direct evidence of culture in the form of pottery, pig and fish bones, and so on. One might initially wonder how a linguist, studying only modern languages whose unwritten ancestral forms remain unknown, could ever figure out whether Austronesians living on Taiwan 6,000 years ago had pigs. The solution is to reconstruct the vocabularies of vanished ancient languages (so-called protolanguages) by comparing vocabularies of modern languages derived from them.
For instance, the words meaning 'sheep' in many languages of the Indo-European language family, distributed from Ireland to India, are quite similar: 'avis,' 'avis,' 'ovis,' 'oveja,' 'ovtsa,' 'owis,' and 'oi' in Lithuanian, Sanskrit, Latin, Spanish, Russian, Greek, and Irish, respectively. (The English 'sheep' is obviously from a different root, but English retains the original root in the word 'ewe.') Comparison of the sound shirts that the various modern Indo-European languages have undergone during their histories suggests that the original form was 'owis' in the ancestral Indo-European language spoken around 6,000 years ago. That unwritten ancestral language is termed Proto-Indo-European.
Evidently, Proto-Indo-Europeans 6,000 years ago had sheep, in agreement with archaeological evidence. Nearly 2,000 other words of their vocabulary can similarly be reconstructed, including words for 'goat,' 'horse,' 'wheel,' 'brother,' and 'eye.' But no Proto-Indo-European word can be reconstructed for 'gun,' which uses different roots in different modern Indo-European languages: 'gun' in English, 'fusil' in French, 'ruzhyo' in Russian, and so on. That shouldn't surprise us: people 6,000 years ago couldn't possibly have had a word for guns, which were invented only within the past 1,000 years. Since there was thus no inherited shared root meaning 'gun,' each Indo-European language had to invent or borrow its own word when guns were finally invented.
Proceeding in the same way, we can compare modern Taiwanese, Philippine, Indonesian, and Polynesian languages to reconstruct a Proto-Aus-

344' GUNS, GERMS, AND STEEL
tronesian language spoken in the distant past. To no one's surprise, that reconstructed Proto-Austronesian language had words with meanings such as 'two,' 'bird,' 'ear,' and 'head louse': of course, Proto-Aus-tronesians could count to 2, knew of birds, and had ears and lice. More interestingly, the reconstructed language had words for 'pig,' 'dog,' and 'rice,' which must therefore have been part of Proto-Austronesian culture. The reconstructed language is full of words indicating a maritime economy, such as 'outrigger canoe,' 'sail,' 'giant clam,' 'octopus,' 'fish trap,' and 'sea turtle.' This linguistic evidence regarding the culture of Proto-Austronesians, wherever and whenever they lived, agrees well with the archaeological evidence regarding the pottery-making, sea-oriented, food-producing people living on Taiwan around 6,000 years ago.
The same procedure can be applied to reconstruct Proto- Malayo-Poly-nesian, the ancestral language spoken by Austronesians after emigrating from Taiwan. Proto-Malayo-Polynesian contains words for many tropical crops like taro, breadfruit, bananas, yams, and coconuts, for which no word can be reconstructed in Proto-Austronesian. Thus, the linguistic evidence suggests