Histogram statistical comparison reference recommendation I am currently wrestling with several pairs of discrete probability distribution histograms (n = 60,000,000+ for each) and are seeking to determine the significance of any and all dissimilarity between the histograms in each pair.
In my research meanderings, I have come across the Bhattarcharyya distance, which is nice and very straight forward to use - but, what is not clear to me is how to measure the statistical significance of the calculated distance. 
Is there a reference that explains the statistical significance of values determined when using the Bhattarcharyya distance?
 A: I seem to have found an answer regarding this. According to the conference paper The Bhattacharyya Metric as an Absolute Similarity
Measure for Frequency Coded Data.
 (Thacker et al. 1997), the significance of the Bhattarcharyya distance:

As the Bhattacharyya measure is equivalent to the Matusita measure we see that the Bhattacharyya metric is
  a chi-squared type statistic in the sense of squared Euclidean distance, although unlike the chi-squared statistic,
  the Bhattacharyya metric measures similarity in a domain where all errors are constant

and specifically, for histograms, the authors propose that the

Bhattacharyya statistic $\sum_{i} \sqrt{R_i} \sqrt{S_i}$ as a measure of similarity between the
  two histograms. For the case of two identical histograms we obtain $\sum_{i} \sqrt{R_i} = 1$ indicating a perfect match.

Where the probability distributions of each of the histograms are $R_i$ and $S_i$
and that

Matusita and Bhattacharyya measures are themselves a form of test statistic but because of their construction
  they require no systematic error correction.

