efficiency, factors which are involved in the comparison of the resistance of the model with 

 that of the ship as deduced from shaft horsepower measurements made on trial-and other 

 quantities besides hull surface finish. ^^ The term "roughness" allowance is therefore very 

 misleading, and for this reason has not been used in the present text. The more rational 

 name "ship correlation allowance" or "factor" has been suggested and the ITTC Presentation 

 Committee has proposed the symbol C^, where the suffix stands for "additional." In Great 

 Britain, where the Froude coefficients are still in general use, the NPL tanks have for some 

 years used a "ship correlation factor" which has different values for different types of shell 

 construction, these values being derived from comparisons of actual ship trial results with 

 corresponding predictions from model tests. The British Admiralty tanks use a similar 

 method of correlation in the form of a quasi-propulsive coefficient factor. The weight of 

 evidence today is that the allowance of +0.0004 above the ATTC line is somewhat too high 

 for modern merchant ships of good welded construction, with a clean, newly painted hull 

 surface, using a standard commercial paint. 



Hadler et al have recently given correlation allowances for 13 merchant ships for 

 which good full-scale and model data were available.^'' They found that the correlation 

 allowances decreased in magnitude both with increase of length of ship and with the date 

 of construction, but were unable to disentangle the relative importance of these two effects 

 because of the small number of ships available. The newer the ship, the better the probable 

 finish of hull surface, but also the newer ships in general are longer. When they considered 

 only seven ships built since World War II, they found the average values of C ^ were +0.00015 

 for the ATTC line correlation and +0.00020 for the ITTC. This difference is small, and will 

 be so for correlations carried out from experiments with large models of the order of 20 ft in 

 length. Because of the divergence of the ATTC and ITTC lines at low Reynolds numbers, 

 the differences in correlation allowances for the two methods will increase with the use of 

 smaller models. 



A large number of ship trials have been correlated with model experiments in Great 

 Britain, ^^ The trials were carried out on some 69 single-screw and 21 twin-screw ships by 

 the British Ship Research Association and the models were run in the NPL tanks. There 

 was some variation in C ^ with speed, but no length effect was obvious. For all-welded hulls, 

 the average value of C ^ using the ITTC line was +0.00015 (with a total scatter of 0.0004) 

 and for half-welded hulls (generally welded butts and riveted seams) +0.0004 (with a total 

 scatter of 0.0007). Using the ATTC line, the corresponding C ^ values were +0.00005 and 

 + 0.0003. Clements pointed out in this paper that the best results achieved to date corre- 

 spond to a correlation allowance C ^ of zero. 



More recently, results of British ship trials embracing modern large tankers have 

 suggested a definite trend towards lower correlation allowances with increasing length, and 

 a proposal to represent this by the straight line 



C^ = 0.00160 - 0.0000023 • Lgp 

 E-3 



