SAS does not have any proc or statement for doing this! you must right a program using SAS matrix language (proc IML) for your data to do this! 

Dear Mr. Oluwaseun

I`m using the IML SAS procedure for calculation variance - co-varance matrix, genotype and phenotype variance - co-variance matrix, co-heritability and phenotype and genotype correlation and genetic parameter. use SAS to calculate variance- co-variance matrix, genotype and phenotype variance - co-variance matrix and co-heritability. and use excel to calculate phenotype and genotype correlation and genetic parameter. if you want, see my paper [Evaluation of genetic parameters in agro-physiological traits of wheat (Triticum aestivum L.) under irrigated condition]

Article Evaluation of genetic parameters in agro-physiological trait...

If the answers you have already got do not help you, try searching the past issues of Crop Science and Agronomy Journal where SAS code may have been provided for estimation of  phenotypic, genotypic and environment coefficient correlation coefficients.  

When we get correlation from raw data we call it phenotypic correlation; when we remove the environment error, then the correlation (on means) among means may be called as genotypic correlation...

That doesn't appear to be quite right Luis! What one first needs are variance-covariance components for the two traits of interest, using eg ReML at both phenotypic and genotypic levels. These are then used to estimate the phenotypic and genotypic correlations.  

Subhash, your process is a more sophisticated (and accurate) form of isolating different kinds of errors. The bottom line is removing extraneous errors from means where the programs you cited do. I prefer proc iml of SAS, the one I know.. Remember that was not an easy task 30 years back...

Yes Luis.

the genotypic correlation coefficient is based on expected value of the used design such as RB or Split-Plot.... so it would need the raw data and it couldn't be carried out on means! because the expected values of mean squares just could be calculated using the raw data with! This is also true for enviromental variance and covaraince and correlation....

thank you, all. am very grateful. but does anyone have prototype of this statement for convenience and accuracy.

HI Omikunle, I've been collecting genetic SAS routines around the internet based on paper's, but I've been developing some routines where's complicated to get easily on inernet. First of all you must tell us what's your design eg. RCBD and if you're considering 1 environment or Multi-Environment Trial assay...Maybe I can post here your script (If I have). Take Care.

hi Dr. Silva. am working with maize using diallel analysis, with lattice design and a multilocational trial. thanks