178 Professor (J, Piazzi Smyth's 



rally about I to f inch in diameter, and the tubes^ which have the 

 divisions etched on the glass, are in the longest cases 18 inches in 

 length. These he calls generating thermometers ; and as no ther- 

 mometer is constant in wide ranges, and at different times, he has 

 other thermometers, smaller, and with only a portion of the scale from 

 freezing to boiling ; and prefers to use a special instrument for a 

 special temperature. 



The first practical difficulty is the equality of the bore of the 

 tube : round bores are most even, but flat ones are easy to read off. 

 In either case, care should be taken to pick out one with no sudden 

 irregularities. To examine into the quantity of these, portions of 

 the columns must be broken, and passed along the tube, first with 

 one length of column, bisecting the extent of the scale, then trisect- 

 ing, and so on, until every tenth or every fifth division has been 

 tested. A little skill is required to continue to break off portions 

 of the column without the application of a lamp to the bulb, as has 

 previously been practised ; and consists in dexterously reinverting 

 the thermometer, which has just been turned downwards, and so bring 

 the vacuum bubble, which will thereby have been formed in the 

 bulb, to the neck of the tube. 



Exceeding pains are then taken, in determining the freezing and 

 boiling points, that the pounded ice in the one case, and the boiling 

 water in the other, are neither above nor below their natural tempe- 

 ratures ; but even when this is accomplished there arises the greater 

 difficulty still, that heat seems to produce two sorts of expansions 

 on bodies, viz., the ordinary, or that which ceases with the increase 

 of temperature, and may be called the periodical expansion ; and 

 another, which may be called a secular expansion, much smaller in 

 amount than the other, but continuing to exist days, or weeks, or 

 months, according to the bulk of the article acted on. 



For strict purposes, therefore, to determine correctly temperatures 

 below boiling, the thermometer should be boiled just previous to 

 being used. But to avoid this trouble, Mr Sheepshanks prefers, for 

 noting ordinary climate temperatures, to have a thermometer 

 marking only to about 70° to 80° Fahr., compared from time to time 

 with one of the generating thermometers, and having its freezing 

 point occasionally tested. Thus avoiding in any one instrument the 

 application of any exceeding variations of temperature. 



This secular change probably affects the glass of the bulb more 

 than the mercury contained therein, and is equally present in every 

 metal, and indeed all substances, greatly to the confusion of our at- 

 tempts to construct standard scales for exact scientific purposes. The 

 age of the bar, and its former experiences, now seem to be elements 

 in determining its length, in addition to the temperature at the 

 moment. 



The old custom of making measuring-rods of wood was neces- 

 sarily exploded before the advancing requirements for accuracy. 



