Smith B H
Museum of Anthropology, University of Michigan, Ann Arbor 48109.
Am J Phys Anthropol. 1994 Jul;94(3):307-25. doi: 10.1002/ajpa.1330940303.
Smith ([1986] Nature 323:327-330) distinguished patterns of development of teeth of juvenile fossil hominids as being "more like humans" or "more like apes" based on statistical similarity to group standards. Here, this central tendency discrimination (CTD) is tested for its ability to recognize ape and human patterns of dental development in 789 subadult hominoids. Tooth development of a modern human sample (665 black southern Africans) was scored entirely by an outside investigator; pongid and fossil hominid samples (59 Pan, 50 Gorilla, and 14 fossil hominids) were scored by the author. The claim of Lampl et al. ([1993] Am. J. Phys. Anthropol. 90:113-127) that Smith's 1986 method succeeds in only 8% of human cases was not sustained. Figures for overall success of classification (87% humans, 68% apes) mask important effects of teeth sampled and age class. For humans, the power of CTD varied between 53% and 92% depending on the number and kind of teeth available--nearly that of a coin toss when data described only two nearby teeth, but quite successful with more teeth or distant teeth. For apes, only age class affected accuracy: "Infant" apes (M1 development < or = root cleft complete, unemerged) were usually classed as humans, probably because the present developmental standard for great apes is in substantial error under 3 years of age. "Juvenile" apes (M1 > or = root 1/4), however, were correctly discriminated in 87% of cases. Overall, CTD can be considered reliable (accuracy of 92% for humans and 88% for apes) when data contrast development of distant dental fields and subjects are juveniles (not infants). Restricting analysis of fossils to specimens satisfying these criteria, patterns of dental development of gracile australopithecines and Homo habilis remain classified with African apes. Those of Homo erectus and Neanderthals are classified with humans, suggesting that patterns of growth evolved substantially in the Hominidae. To standardize future research, the computer program that operationalizes CTD is now available.
史密斯（[1986]《自然》323:327 - 330）根据与群体标准的统计相似性，将幼年化石原始人类牙齿的发育模式区分为“更像人类”或“更像猿类”。在此，对这种中心趋势判别法（CTD）识别789只亚成年类人猿牙齿发育的猿类和人类模式的能力进行了测试。一个现代人类样本（665名南非黑人）的牙齿发育情况完全由一名外部研究者评分；猩猩类和化石原始人类样本（59只黑猩猩、50只大猩猩和14个化石原始人类）由作者评分。兰普尔等人（[1993]《美国体质人类学杂志》90:113 - 127）声称史密斯1986年的方法在人类案例中成功率仅为8%，这一说法并未得到证实。分类总体成功率的数据（人类为87%，猿类为68%）掩盖了所采样牙齿和年龄组的重要影响。对于人类，CTD的效力在53%至92%之间变化，具体取决于可用牙齿的数量和种类——当数据仅描述两颗相邻牙齿时，几乎和抛硬币的概率一样，但牙齿数量更多或距离更远时则相当成功。对于猿类，只有年龄组影响准确性：“幼年”猿类（M1发育≤牙根裂开完全，未萌出）通常被归类为人类，可能是因为目前大猩猩的发育标准在3岁以下存在重大误差。然而，“少年”猿类（M1≥牙根的1/4）在87%的案例中被正确区分。总体而言，当数据对比远处牙区的发育情况且研究对象为少年（而非婴儿）时，CTD可被认为是可靠的（人类准确率为92%，猿类为88%）。将化石分析限制在满足这些标准的标本上，纤细南方古猿和能人牙齿发育模式仍被归类为非洲猿类。直立人和尼安德特人的则被归类为人类，这表明生长模式在人科中发生了显著演变。为规范未来研究，现已提供实施CTD的计算机程序。
