Austin-Eastman — Heat Conductivity in ^Yoods. 541 



vice was first used by C. G. Stangel in his work on heat con- 

 ductivity to which reference has already been made. 



It was found that there were two chief sources of en^or to be 

 guarded against, the first of Avhich is the direct effect on the 

 wax of radiation from the source of heat. To prevent this the 

 metal from which heat is communicated to the wood must not 

 be allowed to extend above the edge of the wood, this being much 

 more effective than any system of screens used with a larger 

 heating plate. The second possible source of error lies in the 

 difficulty in obtaining a perfectly good contact between the 

 specimens. To insure this as far as possible, two strips of tin- 

 foil were placed between and the blocks pressed tightly together 

 by means of wedges. It was also found necessary in order to 

 preserve the contact, to plane the blocks frequently as they be- 

 came warped by the heat. With these precautions very uni- 

 form results were obtained. 



In our experiments only the conductivities parallel to the 

 fibers were tested. The piece next to the source of heat (posi- 

 tion 1 in the figure) was in all cases the same specimen of white 

 oak, the others being placed in position for comparison with it. 

 The densities were determined partly by weighing and measur- 

 ing the specimens and partly by the method of immersion. Five 

 sets of angles were taken for each wood and the average of these 

 used in computing the conductivity. These measurements were 

 estimated to be correct to within two per cent. Some errors are 

 undoubtedly introduced into the results by the fact that in the 

 second specimen the flow of heat is not in general strictly paral- 

 lel to the fibers, and since the conductivity at right angles to 

 the fibers is smaller, the true values of the conductivities of 

 some of the lighter woods may be slightly greater than those 

 given. In order to reduce the relative conductivities at least 

 approximately to absolute units, the value for walnut parallel 

 to the fibers, found bv Peclet^ is assumed to be correct for our 

 specimen and the conductivities of all the other woods are ex- 

 pressed in the same terms. 



iC. G. S. System of Units, Everett, p. 128. 



