The Station fire is indeed a great example of the utility of models. By changing the inputs to the model, the talented folks at NIST were able to get their computer program to very closely reproduce what had already been captured on videotape.

The answers to this one were in the back of the book. Sure, once the model was made to match the known characteristics of the fire, it could be queried for more detail than could be gotten from the video, and those details may actually be a close approximation of what really happened.

The difficuly in using models for fire investigation comes about when there is no videotape available, and the model is asked to provide the data. As with the Station fire, the inputs can be changed to arrive at almost any desired result. If the fire spreads too slowly or too rapidly to match an eyewitness's perception, then the HRR of the USER SPECIFIED FIRE can be modified. Of course if the eyewitness is also a suspect, the model can be used to show that the witness's statements about what happened are not true.

We are at an interesting juncture. A valuable tool is available, but care must be taken to assure that it is used properly. As with any computer output, a reality check is required.

Computer models have been used for many years to reconstruct car wrecks, a complicated business, but nowhere close to the complexity of a fire. Anyomne familiar with these models will tell you that the most important component is the one between the chair and the keyboard.