difficulties with Wilcoxon test

["Roland Knapp" <knapp@lifesci.ucsb.edu>]

I have a dataset made up of two columns, one a factor variable with two levels and the second a continuous variable ("days").  I want to evaluate the magnitude of differences in column 2 with column 1 as a grouping variable using a Wilcoxon rank sum test.  The following code

 

wilcox.test(days[ssealii=="0"], days[ssealii=="1"])

produces this result: 

data: days[ssealii == "0"] and days[ssealii == "1"]

rank-sum normal statistic with correction Z = NA, p-value = NA

alternative hypothesis: true mu is not equal to 0

I am puzzled by the NAs for the Z value and P value, as I have run similar analyses many times before on similar datasets.  In addition, if I use a t-test instead of the Wilcoxon rank sum test, I get t values and P values.  The dataset consists of 2650 observations, and if I trim the dataset back in increments of 10 observations, the Wilcoxon test produces Z values and P values once I've reduced its size to 2630 observations. 

Thanks in advance for any insights into how to solve this one.  Roland Knapp