Suppose I have the regression

$\displaystyle Y=X_1 \beta_1 + X_2 \beta_2 +X_3 \beta_3 + X_4 \beta_4+U $

Where each X is an n-vector and each beta is a scalar. Unde the usual assumptions for OLS, Step by step, how can I construct a 95% CI for $\displaystyle \beta_1+\beta_3 $