Whenever you're ready, walk me through how you'd approach this.
If you regress y on x1 and then x1 on y, what is the relationship between the two beta coefficients? How do the R-squared values compare?
Assume both regressions are ordinary least squares with an intercept, one regressor, and nonzero sample variance in both variables. Derive the relationship algebraically and state how the result changes if the intercept is omitted or additional regressors are included.