Thanks for alerting me to this @IFTRM from Honduras. What was going wrong was that the code in Harrell's book uses the princomp() function from the stats package, *not* the princmp() function from the Hmisc package; strange given that the author of the book also created the Hmisc package. In any case, simple changing `princomp()` to `princmp()` meant that the downstream plot() function now allows the k= argument.

So

plot(prin.raw, 
     type= 'lines', 
     main= ' ', 
     k=13,
     ylim=c(0,3))

which now results in what I want.