A DESCRIPTION OF THE PROBLEM :
The last two sentences of this javadoc read like this:

"Note that if the result of this method is passed to the string constructor, only the numerical value of this BigDecimal will necessarily be recovered; the representation of the new BigDecimal may have a different scale. In particular, if this BigDecimal has a positive scale, the string resulting from this method will have a scale of zero when processed by the string constructor."

This makes no sense because it would be an egregious bug if a BigDecimal constructed from a parsed String had a scale zero while the number represented by the original object had a positive scale. The last sentence should say "negative", instead of  "positive". The method's source corroborates this:

    public String toPlainString() {
	BigDecimal bd = this;
	if (bd.scale < 0)
	    bd = bd.setScale(0);
.........

EXPECTED VERSUS ACTUAL BEHAVIOR :
EXPECTED -
In particular, if this BigDecimal has a positive scale, the string resulting from this method will have a scale of zero when processed by the string constructor.
ACTUAL -
In particular, if this BigDecimal has a negative scale, the string resulting from this method will have a scale of zero when processed by the string constructor.

URL OF FAULTY DOCUMENTATION :
http://java.sun.com/j2se/1.5.0/docs/api/java/math/BigDecimal.html#toPlainString()