🚀 go-pugleaf

I am having some problems with nroff string sizes.  They seem to be rather
inconsistent.

In a macro I use a definition of the form
	.ds>1 \\$1
Use of the \w width function says that $1 is 192 units wide (correct as
the string in question is 8 characters long) but that >1 is now 576 units
(24 chars) wide!! Why is it not the same size as $1 ? (The font *is* Roman 
and it does get printed without overstrikes.)

That seems like a simple question but it gets worse.  Later on, just before
>1 gets referenced, nroff still thinks it is 576 units wide. It is 
referenced in a string like this in another macro (for debugging purposes)
	some text \\kx\\*(>1\\ky some more text
Before interpolating the string the horizontal line position is 480 units (x)
and afterwards it is at 1320 units (y); 264 units more than the declared
width of >1.  Finally it now says that >1 is of size 1320 units! Again at
the time these measurments and interpolation are made the font is Roman.

Any help would be appreciated.  If replies are of sufficient interest I
will post them to the net.

--------
Alan Young.    International Computers Limited, Bracknell, UK
UUCP: ...ukc!{stc,reading}!iclbra!awy