322 NINETEENTH CENTURY. FT. III. 



burning certain substances. For instance, if you put com- 

 mon salt in a spirit-flame, it will burn with a yellow colour, 

 while a substance called nitrate of strontium will give a 

 brilliant red flame, and is used in making red fire for the 

 theatres. Many other metals and earths, however, tinge the 

 flame so slightly that you cannot see the colour, and it is 

 only by passing the light through a prism that you can 

 detect it. 



It had long been known that light from white-hot solids 

 when thrown on a prism produces a continuous spectrum, 

 that is, a coloured band unbroken by any dark lines. A 

 white-hot poker, for example, will give the spectrum No. i, 

 Plate I., and so will burning parafiin, because it contains 

 solid atoms of carbon. But burning gases or vapours do not 

 give a continuous band of colour, they only produce a few 

 bright lines, such as those in Nos. 3 and 4. You can see 

 this by looking at an ordinary gas flame through Browning's 

 little spectroscope. 



Now there is a remarkable peculiarity about these bright 

 lines formed by gases or vapours, namely that they are 

 different for the gas or vapour of every different substance. 

 Thus, if you burn any substance containing sodium, a bright 

 yellow stripe will appear as in No. 3 ; while hydrogen will 

 give one red, one blue, and one violet stripe, as in No. 4. 

 This test is so true and delicate that the eighteen-millionth 

 part of a grain of sodium will give the yellow line ; nor does 

 it matter if you bum many substances together, for theM^ 

 vapour of each one will give its own lines without interfering^} 

 with the others. 



It was Sir John Herschel in 1822 who first suggested 

 that by burning substances in a flame, and marking the bright 

 lines which they produced, it would be possible to detect the 



I 



