A book worth considering for those who want a summary of blackboard architectures, and who may also be considering the design of a new expert system using these techniques. It gives a history of previous uses, al though this is neither as detailed nor as comprehensive as that of Englemore & Morgan. The main purpose of the Cassandra book is to suggest an improved architecture for the modular organization of complex systems. Two examples are given, so as a bonus it gives an insight into the complexi ty of Air Traffic Control. However the claims made in the blurb on the inside of the front cover are not really justified, in that it is not proven in the book that the expected advantages of the proposed architecture are realized in practice, since the two examples given as evidence are too small to show the advantages claimed. The book is marred by a number of presentational errors and omissions, the worst is that Figure 2.6, which would be very useful to help in the explanat ion of 'planes ' , is referred to but not actually included. Figures 2.5 and 2.7 are there all right, so this is not just a case of a typing error in the reference. Finding your way around the book is not easy, unless you are prepared to read it straight through, and even then references back are difficult to locate. There is no index to help this, and the Glossary covers Air Traffic Control terms o n l y but not all of them. For instance it is necessary to understand 'quadrantal or semicircular rules ' which are 'referred to above' on p. 257; however there is no explanation of these terms in the Glossary and it took me a lot of searching to find the explanation of them way back on p. 175. I found that I got on better by abandoning any at tempt to read the book straight through. From this experience I would recommend turning to the part on Air Traffic Control first, this gives a background for a study of the examples, then going back to read the descriptions of the architectures. Again the presentation lets the material down, the