60 FRAGMENTS OF SCIENCE. 



results which it is impossible to analyze. It is by a kind of 

 inspiration that we rise from the wise and sedulous con- 

 templation of facts to the principles on which they depend. 

 The mind is, as it were, a photographic plate, which is 

 gradually cleansed by the effort to think rightly, and which 

 when so cleansed, and not before, receives impressions from 

 the light of truth. This passage from facts to principles is 

 called induction, which in its highest form is inspiration ; 

 but, to make it sure, the inward sight must be shown 

 to be in accordance with outward fact. To prove or dis- 

 prove the induction, we must resort to deduction and ex- 

 periment. 



Torricelli reasoned thus : If a column of water thirty- 

 two feet high holds the pressure of the atmosphere in 

 equilibrium, a shorter column of a heavier liquid ought to 

 do the same. Now, mercury is thirteen times heavier than 

 water ; hence, if my induction be correct, the atmosphere 

 ought to be able to sustain only thirty inches of mercury. 

 Here, then, is a deduction which can be immediately sub- 

 mitted to experiment. Torricelli took a glass tube a yard 

 or so in length, closed at one end and open at the other, 

 and filling it with mercury, he stopped the open end with 

 his thumb, and inverted it in a basin filled with the liquid 

 metal. One can imagine the feeling with which Torricelli 

 removed his thumb,, and the delight he experienced when 

 he found that his thought had forestalled a fact never before 

 revealed to human eyes. The column sank, but ceased to 

 sink at a height of thirty inches, leaving the Torricellian 

 vacuum overhead. From that hour the theory of the pump 

 was established. 



The celebrated % Pascal followed Torricelli with a still 

 further deduction. He reasoned thus : If the mercurial 

 column be supported by the atmosphere, the higher we 

 ascend in the air the lower the column ought to sink, for 

 the less will be the weight of the air overhead. He ascend- 



