Hi, i have this method here that approximates square roots. I was wondering what ppl thought about it and whether it is following a correct algorithm. It seems to work good and gave pretty close approximations for all my test casses. The only issue is on the very first call to the function, if 1 is passed in as a guess it returns bogus numbers.

 * Newtons Method for Approx. square roots.
 * Accepts 2 doubles.
 * Argument 1 is the number to take the square root of.
 * Argument 2 is the approximation/guess of the square root.
 * x =  1/2( x + A/x)  <-- Formula being used.
double approxSQRT(double A, double approx)
    double previous=approx;  

    approx = (0.5)*(approx + ( A/approx));

    //When sequential approximations have a negligable difference the
    //   approximation should be very close to the actual square root.
    if(previous - approx < 0.0000001 )

    return( approxSQRT(A,approx));