there is implied in order to achieve convective mixing to a depth h, 

 an ice equivalent 



r r o- + (h)+0.087} i 



* 1, (h).= h [161.88 {- L ^ j-130.72] (22) 



where the numerical units of h, CT (t), S Q ^ and l^(h) are defined as 

 they were for (10). 



D. CONCLUSIONS 



This new approach to the ice potential computation technique has 

 several advantages: (a) it is simpler and speedier than the regular 

 method; (b) one can see immediately, by plotting the points (^h,^ ^) 

 whether freezing or freezing with subsequent melting occurs as a result 

 of thermohaline convection; and (c) it is easy to estimate the depth of 

 mixing for which freezing initially occurs. When the curve, generated 

 by the points ( °"h, S h ) , remains parallel to a constant percentage 

 line (constant l-j/h^'or it turns back toward the density of the freezing 

 line, tremendous quantitites of thermal energy are involved for addi- 

 tional increases in the ice thickness,, One can decide what depths 

 of mixing are of interest (usually that of the initial freeze and/or 

 that for some specified ice thickness) and compute the associated Q'S 

 (sensible heat losses) alone by the resultant formula proved in 

 APPENDIX I. 



" See Appendix III 

 -sb;- For plotting lines of constant \*/h on either "Ice Determination 

 Graph", the following relation is obvious: 



a- h =cotoc j [(.9k)(l. | /h)+l] S 0jh + bj . (23) 



