644 



NATURE 



[February 12, 1920 



with four fingers, the Toa representing a four-fingered 

 hand. The Muramura, Wutjulvana, had a servant 

 whose index and middle fingers had partly grown 

 together, as indicated by the Toa. The Toa also has 

 a geographical significance, for when Wutjukana 

 came to a gorge which divided into four branches, 

 one being deeper than the others, he said to himself, 

 'This place loolcs like the hand of the servant,' and 

 so he gave it this name." 



" 66. Witjikurawinpani (Tirari Tribe). — To the 

 tracks of the whirlwind in the sand. When the Mura- 

 mura, Patjalina, once came, hunting, to this place, he 

 noticed that a whirlwind had passed over it, which 

 had effaced the tracks of animals and had swept 

 together a litter of leaves and grass ; hence he named 







Fig. I. iG. 4. 



it thus. The white knob represents a sandhill over- 

 grown with bushes (red dots) which adjoins Cooper's 

 Creek. The crescent-shaped, black figure below indi- 

 cates the creek itself, and the black vertical band a 

 deep waterhole at the foot of the hill, which has been 

 washed out by a flood. The surrounding borders of 

 white and yellow signify soil of these colours, and the 

 white spots trees." 



" 187. Muramurawintini (Tirari Tribe). — To the 

 Muramura 's hair. So named because on this plain 

 the Muramura, Patjalina, tore out his hair and threw 

 it away. The white colour represents the plain, with 



NO. 2624, VOL. 104] 



watercourses (red and yellow stripes), and, in accord- 

 ance with the name, the Toa bears a tuft of hair." 



When a native is about to remove to another camp 

 he makes a Toa representing the locality to which he 

 is removing and sticks its pointed end into the earth 

 of the camp which he is leaving. Signs are made on 

 the ground directing attention to its presence. His 

 friends who arrive later recognise the significance of 

 the Toa, and are thus made aware of the place to 

 which he has gone. 



The whole collection and its elucidation form a 

 most interesting contribution to the study of .Aus- 

 tralian svmbolism. Sidney H. R.\y. 



HUMAN METABOLISM.^ 

 T^HE first of the monographs before us deals with 

 -'■ the prediction of basal metabolism from a know- 

 ledge of individual physical and biological constants. 

 The usual process has been to multiply the subject's 

 surface area, as deduced from the height and weight 

 by du Bois's method, by the average Calorie output 

 per square metre determined from a " standard'' 

 series, it being assumed that the metabolism per unit 

 area in adults is approximately constant. The order 

 of the error involved and the improvement effected 

 by Drs. Harris and Benedict's process can be gauged 

 by the following example. In these authors' series of 

 T36 adult males, the mean twenty-four hours' basal 

 output was 1631-7 Calories with a standard deviation 

 of 2047 Calories; the latter is 12-5 per cent, of the 

 mean. The corresponding mean per square metre of 

 surface was 925-5 Calories with a standard error of 

 74-5 Calories, about 8 per cent, of the mean. 



It follows that if the distribution around the mean 

 be assumed to be normal (actually, using a coarse 

 unit of grouping, we find that the authors' series for 

 women is in good agreement with a normal dis- 

 tribution ; the series for men is somewhat less regular, 

 but not a very improbable sample from a normal 

 population), the assignment of 9255 Calories per 

 square metre as the basal metabolism of an unknown 

 individual is subject to a standard error of ±8 per 

 cent. But if a prediction is ba.sed upon a multiple 

 regression equation of the first degree, the other 

 variables used being height, weight, and age, the 

 standard error of the prediction falls from 204-7 

 Calories to 101-7 Calories, or the average percentage 

 is 6-2. 



These remarks assume that the correlation is normal 

 correlation. As this may not be the case, and as a 

 complete stiidv of the form of the regression cannot 

 profitably be made without a larger collection of 

 data, the authors have empirically tested the accuracies 

 of the several methods, comparing the predictions 

 based on their equations for samples not utilised in 

 the calculation of the constants with those afforded 

 bv the surface rule. In nearly every case the regres- 

 sion equations give results closer to the truth than 

 does the surface rule. 



We mav illustrate with a couple of examples taken 

 at random from the data. A man aged twenty-nine, , 

 weighing 66 kg. and 177 cm. tall, had a basal meta- 

 bolism of 1695 Calories. The surface rule would 

 assign 1675 Calories, with a standard error of 135 ; 

 the multiple regression formula gives 1664 Calories, 

 with a standard error of 102. Here both results 

 are good. In another case a man aged forty-three, 

 weighing 58-5 kg. and 181 cm. tall, had a basal mcta- 



1 (i)"A Rioraetric Stud v of Basal Metaboli-sin in Man." By J. Arthur 

 Harris and Franris G. Benedict. Publication No. 279. Pp. vi + a66. 

 (Washineton : '1 he Carnegie Institution of Wa-h'nston, 1019.) 



(2) " Human Vitality and Efliciencv under Prolonged Restricted Diet." 

 By Francis G. Benedict, Walter R. Miles, Paul Roth, and H. Monmouth 

 Smith. Publicalinn No. 280. Pp. xi-t-70T. (Washington; The Carnegie 

 Institution of Washington, 1919.) 



