weights in Statistics[NonlinearFit] within M10:

what does that do exactly - stupid me can not find from the help
saying "weights = Vector -- Provide weights for the data points"
Does that mean the points are weighted (i e the count multiple)
or does it do (fct(x_i) - approx(x_i) )^2 / weight_i or what?

If i save a sheet in M10 and open it then it shows tildes
dispite having interface( showassumed = 0 ) in my ini file.
In M9.5 that was ok (using WIN ME in both cases, classical
interface of course ... and having asumptions in the sheet)

How this ugly behaviour can be avoided?

pls check the attached worksheet ... any reason for that behaviour?

PS: not attached due to some oddity of the forum software,
"Error attaching file 102_strangeRoot.mws: exceeds maximum file size"

so take it from http://www.axelvogt.de/maplekram/strangeRoot.mws
(or *.zip if perfered, it is below 4 kB anyway ...)

Entering

Tst0:= 0.5, 27.1, 0.1, 0.2, 24.0;
tst0:= beta=Tst0[1], lambda=Tst0[2], sigma=Tst0[3], P=Tst0[4], M=Tst0[5];

someFct([1/2*P, 1/lambda*beta],[(lambda+beta)/lambda],-sigma*(1+z)^lambda);
eval(%,evalf([tst0])); subs(z=0.1, %);

myTerm:=subs(someFct=hypergeom,%);

and using evalf(myTerm) I get .9981665521 while evalhf(myTerm) gives me

Error, remember tables are not supported in evalhf

What happens here?