Beta weights and semipartial correlation


I am a quasi-beginner in the field of multivariate statistics, and have a quasi-beginnerish question about beta weights. I would like to know what is the source of the differences between beta weights and semipartial (part) correlation?

From the logic of the method for obtaining beta weights (regressing a dependent on a residual of an independent controlling for all the other independents) I find it completely puzzling that there even is a difference.

Many textbooks and internet guides refer to beta weights as uniqe contribution to the explanation of the criterion variance. How come, then, the beta weights are almost always greater in magnitude then the corresponding semipartial correlation coefficients?

Any help would be appreciated!