January i6, 1896] 



NATURE 



245 



probably be a difference of temperature between the coils, and 

 the mean would be taken. On testing the matter, however, no 

 such difference was found, even up to the highest temperature 

 used, 250° C. If any difference existed it was less than one-tenth 

 of 1° C. , as that amount could have been measured. This 

 being the case in a narrow tube where the air circulation was 

 hindered by a number of mica discs (see Physical Review, 

 February 1893, for description of apparatus), it is improbable 

 that there is any appreciable difference when there is no hind- 

 rance to the air currents except the heating coil of platinum. 



I have never used the apparatus for very high temperatures, 

 but see no reason why it should not be so used ; and it would 

 apparently present a number of advantages for such work, chief 

 among which is the fact that the temperature would be obtained 

 by direct comparison with an air or nitrogen thermometer, and 

 no assumptions made as to the law of variation of thermo electric 

 force with temperature. 



It can, of course, be made self-recording by placing a recording 

 pressure gauge in place of the mercury column, the observer 

 simply keeping the galvanometer at zero by manipulation of the 

 carbon resistance. 



It is obvious that platinum resistance coils can be used instead 

 of the thermo-junctions, the platinum resistance coils forming 

 two arms of a Wheatstone bridge, and the galvanometer placed 

 across them. 



A method of measuring the heat conductivity and tempera- 

 ture coefficient of metals devised by me, and at present being 

 used by one of my students, may be of interest. A metal bar, 

 well annealed, polished, and with special precautions taken to 

 preserve homogeneity of physical state, has its ends placed in 

 two mercury baths, A and B. The bar is protected from radia- 

 tion by concentric polished metal tubes. A is heated electri- 

 cally, and B cooled by a water tube. In the regular laboratory 

 exercise, I have been in the habit of letting the students use 

 thermometers. But in this case, where accurate results are re- 

 quired, capillary tubes are led off from A and B, filled with 

 mercury, thus forming a thermo-junction. Another thermo- 

 circuit has its junctions placed at the entrance and exit of the 

 cooling water. It would, of course, be possible to measure the 

 thermo-voltage directly by standard cell ; but instead of this, 

 in the present method, the two thermo-circuits are balanced 

 against one another, the elements which are immersed in the 

 water being chosen so as to have a much higher thermo-voltage 

 than the copper-mercury couple. It is seen, without much 

 difficulty, that by this means the conductivity of the copper may 

 be measured without knowing more than one temperature, and 

 that only approximately, to a considerable degree of accuracy. 

 As the experiments are not concluded, I am unable to state 

 definitely what the value of the method is, but the indications 

 are that it will prove successful. Reginald A. Fessenden, 



Western University of Pennsylvania. 



On Crookes' Spectrum of Helium. 

 In his investigation on the spectrum of helium,^ Crookes has 

 examined the spectrum of five different samples of gas, two being 

 developed from cleveite (No. i and No. 5), another from uran- 

 inite (No. 2), and two from broggerite (No. 3 and No. 4). Sample 

 No. 5 has been prepared with special care, and is designated 

 " helium purissimum." The five spectra are by no means 

 identical, and it has been concluded that besides helium there 

 are other gases present. E. A. Hill - has even gone so far as 

 to infer the existence of at least fifteen new elements from the 

 comparison of these five spectra. Thirty of the seventy-nine 

 wave-lengths measured by Crookes coincide (within the limits 

 of error) with wave-lengths that we have measured in the spec- 

 trum of cleveite gas.' But the remaining forty-nine lines, many 

 of which are strong, do not coincide with any of ours. As far 

 as we know, it has not been noticed that thirty-three of these 

 forty-nine lines almost certainly belong to argon, among them 

 nearly all the stronger lines. Six more may also be argon lines, 

 but the identification is rather doubtful. Two<lines in all prob- 

 ability are mercury lines, which naturally are likely to appear in 

 a vacuum-tube made by means of a mercury pump. One line 

 may be due to carbon. The table on p. 246 contains a list of 

 the forty-nine wave- lengths that do not coincide with wave- 



* Chemical News, August 23, 1895. reprinted in Nature, August 29, 

 1895. 

 ■-* American Journal 0/ Science, November 1895. 

 3 Berichte der Berl. Akad., July 1895. See also Natuke, September 26, 



1895, 



lengths that we ascribe to helium, and gives their probable 

 origin. The wave-lengths of argon lines are taken from Kayser 

 (Chemical A^ews, August 30, 1895), Eder and Valenta {Ber. der 

 Wiener Akad., October 24, 1895), and from Crookes' own 

 measurements. 



3890 5 and 3885 9 are strong lines that have been seen in the 

 spectra of all five 'samples. Crookes considers them as satellites 

 or components of the strong line between them, the wave-length 

 of which is 3888785 according to our measurements. But as 

 our photographs show that this line is single, or if not single 

 has a weak component 0*05 lower, which can only be observed 

 with much greater dispersion than Crookes has used, we are 

 inclined to l^elieve that 3890*5 and 3885-9 are spurious lines due 

 to some error of apparatus having made their appearance on 

 account of the enormous energy of 3888*8. 



Of the five remaining lines, three are only of intensity 2. The 

 two stronger ones have only appeared in the gas from uraninite, 

 and may possibly belong to a substance hitherto unknown. But 

 it is far from being established. 



C. RUNGE AND F. PaSCHEN. 



Hannover, Technische Hochschule. 



The Place of " Pithecanthropus " in the Genealogical 

 Tree. 

 In the report on the scientific meeting of the Royal Dublin 

 Society on November 20, in Nature of December 5, 1895, it is 

 stated that I placed Pithecanthropus in the genealogical tree^ 

 drawn by Prof. Cunningham, below the point of divarication of 

 the Anthropoid apes from the human line. This indeed I did. 

 But this statement could be misleading as to my real views on the 

 genealogy of Pithecanthropus, such as I stated them already on 

 p. 38 of my original memoir (^^ Pithecanthropus erect us, Eine 

 menschenahnliche Uebergangsform aus Java," Batavia, 1894), 

 and more fully at the last meeting of the Anthropological 

 Institute of Great Britain and Ireland, on November 25. 



Cercopitheeidae Tfylgbatt 



Homo ^nlhropopi/heciis Gorilla: 



NO. 1368, VOL. 53] 



ProcercvpUJiecus 



It may not be superfluous to explain my views here by means 

 of the accompanying diagram, representing the evolution of the 

 Old World apes from a hypothetical common ancestor, whom 

 I call Procercopithecus. 



In Prof. Cunningham's tree, figured in Nature of December 

 5, p. 116, he regards the left branch as all human, the 

 right one as entirely simian, and he placed Pithecanthropus 

 midway between recent Man and the point of divarication. 



