By treating as independent variables the various electrical properties 

 at the given exposure periods, prediction equations for atmospheric performance 

 were established. The absolute values of the correlation coefficients (the square 

 roots of the "R^" values) of the best linear prediction equations are shown in 

 Table 9. These correlation coefficients are much larger than those obtained for 

 the individual electrical properties. It appears likely that still higher correlation 

 coefficients can be obtained by developing prediction equations which employ 

 as independent variables not only the electrical properties of the coatings but 

 also other properties, such as permeability. 



It had been claimed that coatings with the very low dissipation factors, 

 and with dissipation-factor-versus-frequency curves of low slopes, will perform 

 better.^-'' 



The dissipation factors obtained at different frequencies for six of the 

 systems are shown in Figure 20. These are the systems which had the lowest 

 curve and the highest (Systems 1 22 and 1 20), the curve with the least slope 

 and the greatest (Systems 121 and 111), and the systems with the best and the 

 poorest performance in atmospheric exposure of unscribed panels (Systems 1 16 

 and 121). The curves show values obtained after approximately 2 weeks of 

 exposure in seawater at 25°C. The dissipation factor curves obtained after 

 approximately 2 to 3 hours may be quite similar to or may deviate from those 

 obtained after approximately 2 weeks. The system that showed the greatest 

 variation in this time interval was System 1 19, and both curves for this system 

 are shown in Figure 20. 



When 12 of the coating systems are arranged according to the increasing 

 quality of their dissipation factor curves (which are shown in Figures 20 and 21), 

 the following approximate order results: 122, 116, 121, 117, 112, 118, 113, 

 114, 111, 123, 119, 120. This order bears little resemblence to the order in 

 Table 7, or to the orders in Tables 6 or 8. 



It has been suggested that changes in electrical properties occur much 

 sooner than visual changes in performance, and that electrical measurements 

 may therefore be used to detect early failure of coatings."* In the present series 

 of experiments, only one of the systems allowed noticeable rusting, and five 

 systems had detectable blistering after 2 years of exposure in seawater at 25°C. 

 All these five systems showing blistering had strong water uptake in the first 

 10 days (as indicated by AC resistance changes), but six of the remaining eight 

 systems also had strong water uptake. Three of the five systems showing blistering 

 developed low AC or DC resistances on one or more panels within 1 50 days, 

 whereas only one of the remaining eight systems developed low resistances within 

 this time. This difference appears to be meaningful. However, it does not appear 

 that the correlation between such low resistance and the degree of blistering is 

 sufficiently high to make such nneasurements useful for the prediction of coating 

 performance. 



44 



