One way to proceed would be to allow each piece of software to read the data it needs from its own file. Such a scheme would quickly decline into chaos as the number of data files grows. The alternative is to have one central system read all the required data but then the problem is that this central service would have to be updated each time new titles are introduced. The compromise solution is to have a system that can read completely arbitrary sets of titles and then pass them on demand to any software that requires it. This system is called Management of Titles (MTT). As it actively manages titles, MTT has the ability to solve another problem: time dependent data. Some titles, particularly detector description data, are not constant for all time e.g. PMT calibration data. Such data will be multi-valued, each set associated with a date validity range. A primary function of MTTis to ensure that, for the data being currently processed, the right version of such titles is input even if this means changing versions while processing.