There are a few rules for compiling reference works, and the utility of the publication is woefully compromised if those rules are not followed.
The Oxford Companion to British Railway History (1997) is a great book, and fascinating reading, but seen from the perspective of reference publishing, it leaves a lot to be desired.
Help the user finds entries where they might look
Well, this book has no index. As an alphabetical list of themes (such as “Leeds”, “liveries”, or “Post Office” it works reasonably well. But what if the user’s perception of a topic differs from that of the compilers? What if, for example, the user seeks an entry about “crossing” and finds nothing, only to find there is an entry for “junctions” (and I am intrigued what this entry is plural – why isn’t it “junction”?)
Other inconsistences include “stations, passenger” but “reopened railways”(why not “railways, reopened”?
Don’t use abbreviations that assume knowledge on the part of the audience
There are several references to HSTs, without explaining these are references to the High Speed Train (and there is no entry for HST, and no list of abbreviations. You have to know what HST stands for to look it up!)
By this I mean what some might say is part of the charm of this book. The entry of “junctions” meanders into a charming discussion of stations placed at a junction and how their name is derived – is it the junction for a place, or at a place, or near a place? But all this is not very relevant to a definition of the term “junction”.
Overall, it deserves 10/10 for quirkiness (entries on “angling”, “announcing of trains”, “trees, planting of”, but 4/10 for organisation.