Although I hesitate to venture in where others are (wisely) choosing not to
go, it would seem to me that both Brock/Groff and the Gilbert/Nelson
hypotheses have some very useful aspects.

1) Gilbert/Nelson is useful in thinking about the total population of
cars that "should" traverse the total railroad being modeled and is thus a
better than random way to address the proper mix of foreign power on the
road;

2) Brock/Groff is useful in thinking about the particular consist
anticipated for a specific train or category of trains (e.g., peddler
freight, bridge through, etc.).

The corollary would seem to be that while individual trains should be
selected using the Brock/Groff hypotheses (e.g., where did it come
from/where is it going, what do the wheel reports tell us?), the cars used
to fill the trains should be selected from the Gilbert/Nelson "pool" and, if
enough of the trains on the road were modeled, the Brock/Groff and
Gilbert/Nelson hypotheses should tend to converge.

That said, thanks to those of you on the list that suggested some resources
for the 1880s-1930s period. Very helpful!

Eric Hiser
Phoenix, AZ