TIDE-PREDICTING MACHINE NO. 2. H 



the 24 components given in Table B. The summing mechanism of 

 the first machine was improved upon in this, at the suggestion of Sir 

 William Thomson, by giving true Hnear harmonic motion to its 

 elements. It is generally known under the name of the Roberts 

 tide-predicting machine or British tide predictor No. 2. (See plate 

 2.) A description and drawings were pubUshed in The Engineer 

 (London) of December 19, 1879, later in Volume XVI, pages 331-335, 

 of the Great Trigonometrical Survey of India, and also in the sup- 

 plement of the Scientific American, January 23, 1904. 



It was used exclusively at first for the prediction of the tides for 

 India, but later was used to predict tides for all the British Empire, 

 with the exception of Great Britain and Ireland, the tidal predictions 

 for wliich are still made from old tables. This tide predictor No. 2 

 was placed under the immediate charge of E. Roberts until the year 

 1903, when it was removed to the National Physical Laboratory. 



BRITISH TIDE PREDICTOR NO. 3 (nOW IN FRANCE). 



Some time later a third machine, designed by Sir WiUiam Thomson, 

 was constructed by Kelvin & James White (Ltd.), Glasgow. It was 

 fitted with 15 components (see Table B) and was known as British 

 tide predictor No. 3. Greater rigidity than was attained with the 

 conical gearing employed hi the former machines seems to have been 

 considered desirable by the designer, because spur wheels on short 

 and strong shafts in place of conical gears were used for driving the 

 components. It was described briefly in Thomson's Popular Lec- 

 tures and Addresses, volume 3, pages 184-187, and also in the sup- 

 plement of the Scientific American of January 23, 1904. 



This machine, after being used a number of years by Lord Kelvin 

 for experimental work, was acquired by Kelvin & James White 

 (Ltd.), who generally overhauled it and added the component T^ 

 (see Table B). From this firm it went into the possession, about 

 1900, of the hydrographic service of the French Government for use 

 in the preparation of the French tide tables. (See plate 3.) 



BRITISH TIDE PREDICTOR NO, 4 (nOW IN BRAZIL). 



About 1909 or 1910 Kelvin & James White (Ltd.), constructed 

 another machine upon the same model for the BraziKan Govern- 

 ment. It differs from the French machine only in the number of 

 components, of which 12 only are provided. (See Table B.) 



BRITISH TIDE PREDICTOR NO. 5 OR ROBERTS' UNIVERSAL TIDE PRE- 

 DICTOR. 



During the preparation of this paper, information was received 

 from Edward Roberts of the existence of another tide-predicting 

 machine, designed by him in 1906 for his own private use and exhib- 



