Most genealogical books follow the general form that FTM uses for the two descendant GenRpts (NGS and NEHGS). In each, the individuals are assigned numbers, usually printed to the left of their name. You will also see that there are two simultaneous sequences going on: 

1. The main entries, which may be exdented, boldface, or otherwise distinguished. Here we usually find a  précis of the individual's life, including marriages, &c &c. Following main entry will be: 
2. all the offspring entries, grouped by spouse. Here will be the barest facts about the individual. Usually, there will be some indication if there will be found a main entry further along, in the main entry sequence. (Children with no offspring are not usually carried forward.)

Each individual may appear twice: first as a child, then as a main entry, each time with the same assigned number. 

In the Hardcopy Era, printing a book was not something easily repeated. Changes waited for another generation or two to make it worthwhile. Consequently, there were relatively few genealogies for any one family, and one could speak of "Wilson's #160" as a piece of information uniquely identifying one Benjamin Sands (1762-1841). 

In the Computer Era, I can produce dozens GenRpts of my Sands document, each different, and each possibly with different numbers for poor Benjie. These numbers have lost their utility, except within the confines of the individual document, where it could be used, for example, in place of page numbers in the index. (FTM does not do this, of course.) 

A properly defined and implemented Reference Number should again allow unique reference to individuals beyond the individual document in a way that transcends revisions. One could, for example, refer to individuals in one document from another, without worrying that next weeks revisions will upset the number references. 

That is exactly how I use the Reference Numbers. I have split my total database into separate surname/family files. Each file has a unique 4 letter prefix that I use with the AutoRef numbers. 

I also have one Ancestor file, which is an Ahnentafel based on myself. All of my direct ancestors have a unique number by virtue of this file. All others (cousins) have the RefNo assigned by FTM.

My roots are in early 17^th century New England, where intermarriage was common. An intermarriage is represented in my system by duplicating the two partners in each other's files, but carrying their own RefNo (prefixed by an additional asterisk). Children are carried in the father's file only. Each Family File has a Book of standard format. One of the chapters, is a kind of Report entitled "Inter-family Alliances". In it I list all the individuals with an asterisk in their RefNo, with these fields: 

Spouse; Marriage date; Marriage place; Name; RefNo; Ahnentafel