3-t KE1>0KT — 1877. 



observation, and, after a careful examination of each single observational quantity, 

 and of each constant entering into the computations, the writer came to the con- 

 clusion that no error of the magnitude of that in question could have its source 

 anywhere but in connexion with the induction corrections. The values obtained 

 for the horizontal force were (in British units of force) : — 



With magnetometer No. 17. With magnetometer No. 23. 



8-0701 1 8-0841 1 



8-0G98 \ 8-0823 ( 



8-0702 i 8-0916 i 



8-0764 ( 8-0945 ( 



8-0694 ( 8-0965 ( 



8-0707 f 8-0904 ( 



8-0757 1 8-0844 1 



80756 \ 8-0821 ( 



Mean = 8-0730 8-0858 I 



8-0905 1 



8-0902 I 

 8-0858 | 

 8-0905 ( 

 8-0880 f 



Mean = 8-0884 



No. 23 showing an excess over No. 17 of -0154 British units of force, or of -0019 of 

 the whole horizontal force. 



We observe that the greatest value given by No. 17 is less than the least value 

 given by No. 23, and infer that the difference between the two means cannot be 

 attributed to probable error of observation, the value of which for a single deter- 

 mination (about -001 of the whole force) is, moreover, much smaller. 



If we now remove the corrections applied for induction to the vibration observa- 

 tions, the mean value yielded by No. 17 becomes -0004 of the whole force, greater 

 than the mean yielded by No. 23. It thus appears that a small correction, such as 

 we have already seen reason to expect, is required for the vibration observation, 

 but (on an average for the two instruments employed) only of about one sixth of 

 the value of that which it is the custom of magneticians to apply ; and as this 

 small quantity scarcely exceeds the probable error of the mean determination of the 

 horizontal force, it is yet premature to attribute it to any definite cause. Whilst, 

 however, the experiments afford no sufficient reason for applying this small cor- 

 rection, they speak very distinctly in favour of no induction correction at all for 

 the vibration observation as against the common practice. 



To show that the error that we have been discussing is not of that minute order 

 that is usually disregarded, we may mention that it would amount, in the case of 

 the unifilar magnetometer used at the Kew Observatory, to about eight times the 

 probable error of an observation. 



On the Mode of stating some Elementary Facts in Electricity. 

 By Prof/G. Carey Foster, F.R.S. 



On the Telephone. By W. H. Pbeece, Memb. Inst. C.E., fyc. 



In the following paper, instruments employed in the transmission of musical 

 sounds are called tone telephones, and those employed in the transmission of the 

 human voice, articulating telephones. 



In the year 1837, Page, an American physicist, discovered that the rapid magne- 

 tization and demagnetization of iron bars produced what he called "galvanic 

 music." Musical notes depend upon the number of vibrations imparted to the air 

 per second. If these exceed sixteen we obtain distinct notes. Hence, if the 

 currents passing through an electromagnet be made and broken more than sixteen 



