The column named REDLEAK contains an 8-character string that specifies the treatment of red leaks in short-wavelength filters. Possible values are 'MEASURED' if the red leak is measured by observations through a filter that isolates the leak; 'BLOCKED' if a copper-sulfate or other blocking filter is used; or 'IGNORED' if the leak is neither measured nor blocked. 'ABSENT' can be used for long-pass filters, like the V of standard UBV. It is the user's responsibility to determine that blocking is adequate, particularly if interference filters and/or red-sensitive detectors are used. An unblocked red leak can produce very large transformation errors.
If the leak is 'MEASURED', additional information is required, because most instruments do not provide a true measurement of the red leak. The additional information is stored in character columns named RLTYPE and MAKER.
Column RLTYPE may contain the values 'CEMENTED' if the ultraviolet and long-pass components are cemented together; 'LOOSE' if the two components are not optically contacted; 'UNKNOWN' if information is not available. If the filters are loose, the two extra air-glass reflections cause excess loss that must be accounted for.
The column named MAKER may take the values 'CORNING' if a Corning or Kopp (successor to Corning) glass is used for the long-pass component of the red-leak filter; 'SCHOTT' if a Schott glass (or other non-Pyrex base glass) is used; or 'UNKNOWN' if information is not available. This information is required because the Pyrex glass used as the base for Corning filters absorbs appreciably at typical red-leak wavelengths; the measured leak must therefore be increased to compensate for the absorption.
Sometimes only the shortest-wavelength band of a system has red-leak problems (e.g., U of UBV, or u of uvby.) However, if silicon detectors or red-sensitive photocathodes are used, blue and even green (``visual'') bands may have red-leak problems, especially if heavily reddened or late-type stars are observed.
Note that the importance of red leaks depends on the photometric system, the filter set, the detector used, and the stars observed. Failure to treat red leaks correctly will produce serious systematic errors, which cannot be ``transformed away'' by any reduction program. Worse yet, incorrect treatment of red leaks can propagate these errors (through incorrect transformations) into data for early-type stars that otherwise have negligible red-leak problems. It is the user's responsibility to be aware of these problems.
For additional information on red leaks, see Shao and Young , besides the very brief treatment on pp. 109 and 184 of Young . Stetson  has illustrated how severe the problem can be when cool objects (light bulbs!) are observed with CCDs. Note his warning: ``Don't be satisfied with statements like `The red leak is negligible'.'' Unfortunately, the important cautionary remarks of Azusienis and Straizys  regarding reddened stars are available only in Russian. They show that the simple correction formula used by Shao and Young  is not correct for heavily reddened stars.
To sum up, one may say that red leaks should be measured whenever they exceed the accuracy desired in the final results - which is more often than you might think.