How to quantify variance explained from PCA?

Hi,

I want to quantify the amount of variance explained by PCA. However, I want to define the PCs using one half of my data, and test it using the other half as follows:

[COEFF,SCORE,latent,tsquare] = princomp(InputMatrix_TrainingData); %InputMatrix is an 8 x 78 matrix. RecomputedScores = InputMatrix_TestData*COEFF; %Output

This works fine for recomputing the scores based upon alternate data, but how do I recompute the amount of variance explained etc. ?

Thanks