Many thanks - that does the trick. It shows that the author of FITLD does not know the difference between AIPS bytes (1/2 word) and real ones. Fixed the test on buffer size and the MNJ will get it next time it runs. ERic