SECT. 6] TURBULENCE 811 



fluctuations which contribute significantly to a particular effect, such as the 

 shearing stress, lie within a limited range of periods. The experimental method 

 may then be designed to cover this range adequately. 



Fluctuations in the measurements made by current meters have long been 

 known but have usually been regarded as undesirable features to be eliminated 

 by averaging. The first attempts to study the fluctuations themselves were 

 probably those of Thorade (1931) in the River Elbe and later (1934) in the 

 Kattegat, using the Rauschelbach current meter, giving records of the speed 

 and direction of the current every 10 sec. The fluctuations which he attributed 

 to turbulence had periods of several minutes. Mosby (1947, 1949) used twelve 

 sets of revolving cups, mounted on a stand at different heights, from 9 cm to 

 249 cm above the bottom, to measure the current in the Alvaerstrommen, near 

 Bergen. Fluctuations with periods in the range 2 to 15 min were found and 

 usually corresponding fluctuations could be identified at all levels. Similar 

 observations were made (Mosby, 1951) on the Viking Bank, in a depth of 100 m, 

 at a distance of 100 km from the nearest coast. 



In order to record short-period turbulence, Doodson (1940) designed a current 

 meter which would respond to fluctuations with periods down to about 1 sec. 

 In the first experiments, with the meter suspended from a boat, fluctuations 

 were recorded but there was some doubt as to whether they were influenced 

 by the rolling of the boat. Clear evidence of the short-period fluctuations 

 associated with the flow of a tidal current was obtained when the Doodson 

 meter was used in a stand on the bottom in the River Mersey (Bowden and 

 Proudman, 1949). The values of u'^, their relation to the mean current U at 

 various heights and the auto-correlations of the u component were determined. 

 In another series of observations, two current meters were used, mounted in 

 the same stand with varying separations, so that the spatial correlations of u 

 could also be investigated (Bowden and Fairbairn, 1952). 



The introduction of an electromagnetic flow-meter (Bowden and Fairbairn, 

 1956) enabled measurements of the vertical component w as well as the longitu- 

 dinal component u to be made, and also extended the range to rather shorter 

 periods. The measuring head of the electromagnetic flow meter produces a 

 local magnetic fleld and the water flowing through the field has an e.m.f. 

 induced in it. The potential gradient set up in the water is measured by two 

 pairs of electrodes, so that two components of the velocity fluctuations can be 

 recorded simultaneously. The over-all diameter of the head is 10 cm, while the 

 frequency response of the equipment is flat for periods above 1 sec and falls to 

 one half at 0.25 sec. 



A series of observations with this apparatus was made in tidal currents off" 

 Anglesey, North Wales, within 2 m of the bottom in depths of water ranging 

 from 12 m to 22 m. There was no measurable vertical gradient of temperature 

 or salinity, so that the measurements corresponded to flow in conditions of 

 neutral stability. For mean currents U in the range 25 to 50 cm/sec, average 

 values of the root-mean-square fluctuations were ('m2)'/2/C/ = o.13, {w^y^^lU = 

 0.066. These values did not vary much with height in the range 50 to 175 cm. 



