Two sets of changes were made both suggested by L. Michel. ASCII tables with data fields that were blank filled were not being handled properly. According to the FITS standards, numeric fields where the FITS table has blanks are to be treated as containing 0. A parsing error was being returned. The getInt, getLong, and getDouble methods in ByteParser were changed to acoommodate this case (getFloat simply calls getDouble). A new exception, PaddingException, which inherits from FitsException has been added. This exception is thrown when an otherwise valid HDU is not properly padded to the next 2880 byte boundary. The exception class has a getTruncatedHDU method which allows the user to get the information in the truncated HDU. In addition to the new class changes were made in BinaryTable, AsciiTable, UndefinedData, ImageData and RandomGroupsData to throw the exception at the appropriate time. The main Fits method was also updated so that when its readHDU() method is being used, the notional header that is given to the truncated HDU in the Data classes is replaced by the actual header. If a user wishes to ignore padding exceptions, then a FITS file may be read using the following idiom: Fits f = new Fits(filename); try { f.read(); } catch (PaddingException e) { f.addHDU(e.getTruncatedHDU()); } Users may also look at the new nom.tam.fits.test.PaddingTester to see a complete example of this idiom.