Suppose we are interested in regressing variable Y (dependent variable) on two explanatory variables: X1 and X2. We are concerned about nonlinearity in the multiple regression so we transform each variable by squaring (SQX1 and SQX2), calculating the square root (SQRTX1 and SQRTX2), taking the reciprocal (RECX1 and RECX2), and taking the natural logarithm (LNX1 and LNX2) for each explanatory variable. Using the correlation matrix below, what is the best form for the variable X2 when Y is the dependent variable?