`Modern Applied Statistics with S-PLUS'
=======================================
by W.N. Venables & B.D. Ripley
==============================

Errata and comments for the first printing of the third edition.  The
book was completed well before S-PLUS 2000 was finalised, so some late
changes there are not reflected in the book.

p. 19 ll.-15,-9  [1] 3 4 2 1 3 4 4

p. 22 l.4  logical columns are not converted to character in S-PLUS
	5.x and 6.x,  although the statement is correct for S-PLUS 3.x,
	4.x and R.

p. 24 l.-16  v <- 2 * x + y + 1

p. 41 l.9   at least >1<.  Arrays can be of dimension 1 (although this
	is not very useful).

p. 41 l.-6 a[3,5,9] should be a[2,5,10]

p. 66 l.-5.  With legend, pch is a single character string concatenating the
	symbols, not a vector. For numerical pch, use the vector argument
	marks. See ?legend for full details and examples on your system:
	R is different here.

p. 67 l.6  Apparently, these escape codes do not work in S-PLUS 2000,
	despite being documented in the manuals for that system.

p. 134 (5.4) 1/b should be 1/(nb)

p. 167	The prediction problems with factors have been corrected
	in S-PLUS 2000 and 6.x, using code we supplied.

p. 170 last line, displayed equation.  For u <= 1 read |u| <= 1,
	similarly read |u| > 1 for u >= 1.

p. 188 l.-6  names(cpus)[2:7]
	Our code omitted to transform chmax: fortunately this makes
	very little difference.

p. 203 l.2 of text.  Replace `z ratio' by `t-value'.

p. 231 display  zeta_k - eta
       immediately below: remove `< zeta_{K+1}'

p. 252	confint.profile.nls needed to be altered for a fit wth just one
	parameter.

p. 263  Our changes to D and make.call have been incorporated into
        S-PLUS 2000, and so are not in the MASS library supplied with
	that system.

p. 285  display.  ... - \alpha . (This is necessary as the smoothers
	used by gam() do not include the constant.)

p. 325  The sum of squares 0.17692 is not a cross-validated estimate of
        prediction accuracy like the others for this dataset.

p. 351 l.8  tp <- factor(Cushings$Type[1:21])
	is used to drop unused levels of the factor.  On current systems

	    tp <- Cushings$Type[1:21, drop=T]

	would be a better alternative, but both should work.

p. 371  third displayed equation. - is needed before H(t).

p. 451  An HTML-based help.start() is available under S-PLUS 5.1 (but
	not 5.0).  A Java-based help.start() is available under S-PLUS 6.0.


Typos
-----

p. v l.1  `Data Analysis Products Division' (no and).

p. vi	Jos\'e Pinheiro

p. 15 l.14 >use< print(.Last.value)

p. 29 l.8 "likke" should be "like"

p. 191 l.1 of text   dif>f<erent

p. 225, 484 Santner not Santer.

p. 226, 475 Cox and Snell (1981) not 1984.

p. 275 l.-1 omit >the< before each

p. 276 l.-3 wher>e<as

p. 284 The Green & Silverman reference is to pp. 37-8.

p. 494	Mallows' not Mallow's.


New books
=========

Two books have appeared recently covering nlme3 and survival5 in depth:

Pinheiro, J. C. and Bates, D. M. (2000) Mixed-Effects Models in S and
   S-PLUS. Springer-Verlag. 0-387-98957-9.

Therneau, T. M. and Grambsch, P. M. (2000) Modeling Survival
   Data. Extending the Cox Model. Springer-Verlag. 0-387-98784-3.


S-PLUS 6.x changes
==================

S-PLUS 6.x has moved to common code-base for Unix and (upcoming) Windows
versions, so one would expect the versions to behave identically
apart from rounding errors are OS-specific functions.


pp. 36, 7 Subdirectory .Data (or .data: no longer _Data) is used on Windows.

Chapter 12.  A later version of survival5 is used, with slightly
	different output.  In particular, summary.survReg now
	has cor=F as the default.

Chapter 13.  Units in rts objects are no longer interpreted, and, for
	example, series of frequency 12 are not longer printed with
	month headers.

	The output from e.g. acf() is now printed by a print method,
	and so differs in detail.

pp. 449-51.  S-PLUS 6.0 for Unix behaves in the same way as 5.1,
        except that the command has reverted to Splus.  If the command
	Splus -g is used, S-PLUS starts up with a Java-based console
	quite similar to Windows.

	There is a Java-based help system (JavaHelp in MathSoft
	documentation) used by help.start(), replacing the HTML-based
	one in 5.1.  As in all variants of help.start(), help requests
	are diverted to that help system when running, and it can be
	shut down by help.off().

p. 451  Using Splus -e and S_CLEDITOR set to emacs allows the arrow
	keys to be used under Unix.

pp. 452-3  S-PLUS 6.0 for Windows behaves in almost exactly the same way
	as 4.x.  It does use .data and .prefs rather than _Data and _Prefs.


Appendix B.  With very minor changes, the description for S-PLUS 2000
	applies to S-PLUS 6.0.