So, I am doing my project on temperature.  I have the autocorrelations for 60 lags, and 5459 data points.  My problem is when I calculate Q I am getting a value of (5459)*sum(autocorrelations^2) = 4700.  From what I can tell, 95% confidence would be < 79.08.  Obviously well out of my range.

My autocorrelations are taken from the stationary process of the deseasonalized data.  No single one is over 2/sqrt(T).  I just need help with what I'm doing wrong.  I don't know if my calculated Q is wrong (my guess) or am I reading the chi-square table wrong?  Both?  Is T not the # of points but rather the number of lags?  Thanks!

[NEAS: See attached PDF file.]

Sorry, I think what I was confusing was the autocorrelation given by the CORREL function to the calculated autocorrelation, corrected for the degrees of freedom.  I wasn't using the CORREL function, but my code was doing the same thing.

I'm trying to follow the steps outlined here, http://www.neas-seminars.com/discussions/download.aspx?id=2133&MessageID=7747

about finding the autocorrelations, Box-Pierce Q, etc, and it all seems to work fine, BUT, it's still giving me a Box-Pierce Q-statistic for 50 lags of 4,569.472.  Does this just say that my de-seasonalized temperatures are NOT a white noise process, and that they follow some sort of ARIMA process?