diff --git a/cols.1 b/cols.1
index 5383e47..10bddb1 100644
--- a/cols.1
+++ b/cols.1
@@ -1,34 +1,40 @@
-.TH COLS 1 sbase\-VERSION
-.SH NAME
-cols \- columnize output
-.SH SYNOPSIS
-.B cols
-.RB [ \-c
-.IR chars ]
-.RI [ file ...]
-.SH DESCRIPTION
-.B cols
+.Dd December 8, 2014
+.Dt COLS 1 sbase\-VERSION
+.Os
+.Sh NAME
+.Nm cols
+.Nd columnize output
+.Sh SYNOPSIS
+.Nm cols
+.Op Fl c Ar chars
+.Op Ar file ...
+.Sh DESCRIPTION
+.Nm
 reads each file in sequence and writes them to stdout,
 in as many vertical columns as will fit in
-.I chars
+.Ar chars
 character columns.
 If no file is given, cols reads from stdin.
-.SH OPTIONS
-.TP
-.BI \-c " chars"
-specifies the maximum number of character columns to use
+.Sh OPTIONS
+.Bl -tag -width xxxxxxxx
+.It Fl c Ar chars
+Specifies the maximum number of character columns to use
 (unless the input contains lines longer than
-.I chars
+.Ar chars
 characters).  By default cols tries to figure out the width
 of the output device, if that fails it defaults to 65
 chars.
-.SH BUGS
+.Sh BUGS
 This implementation of
-.B cols
-assumes that every byte is a character
-which takes up one column on the screen.
-It does not handle TAB characters correctly.
-.B cols
+.Nm
+assumes that each UTF-8 code point occupies one character cell,
+and thus mishandles TAB characters (among others).
+.Pp
+.Nm
 currently mangles files which contain embedded NULs.
-.B cols
-does not allow the user to set a default width in its environment.
+.Sh HISTORY
+.Nm
+is similar to the mc(1) command on Plan 9.  It was renamed to
+.Nm
+to avoid the name collision with the popular file manager
+Midnight Commander.
diff --git a/cols.c b/cols.c
index 9b5546e..cde9777 100644
--- a/cols.c
+++ b/cols.c
@@ -20,7 +20,7 @@ static long n_rows;
 static void
 usage(void)
 {
-	eprintf("usage: %s [-c chars] [FILE...]\n", argv0);
+	eprintf("usage: %s [-c chars] [file ...]\n", argv0);
 }
 
 int