Merge two EOFs (principal components)

I'm doing an EOF analysis to my data, and I make a decision on how many EOFs to retain. I also used the North's rule of thumb (North et al. 1982, Sampling errors in the estimation of empirical orthogonal functions) to see which EOFs can be "separated". The attached file shows how my eigenvalue spectrum looks like. The sampling errors of the first two EOFs overlap, so, according to North I cannot separate the first two EOFs since they both are linear combinations of the true EOFs. However, in terms of data reconstruction, if I take both of them is all right. North says that the important thing is not to define the cut-off level in the middle of those two.

Now, I would like to combine this two EOFs into a single one. Is this possible?

Putting it in a more mathematical way, if I want to reconstruct the data only with the first two EOFs I should do PC1.V1T + PC2.V2T, with PC1 and PC2 being the principal component time series and V1T and V2T the eigenvectors (tansposed) from which we derived the EOFs. I want to know if I can merge those two EOFs into a single one, and if I can, what would the PC of that 'new' EOF look like?

Any help would be truly appreciated!