Floating point overflow

uses crt;
var xn,xk,dx,x,s,e,si,i:real;
    n:integer;
Function f(x:real):real;
begin
f:=0.5*ln((x+1)/(x-1));
end;
begin
clrscr;
repeat
write('vvedite nachalo intervala |x|>1 xn=');
readln(xn);
until xn>1;
repeat
write('vvedite konec intervala x>',xn:0:1,' xk=');
readln(xk);
until (xk>xn)and(xn>1);
repeat
write('vvedite shag tabulacii dx>0  no <',xk-xn:0:1,' dx=');
readln(dx);
until (dx>0)and(dx<xk-xn);
write('vvedite tochnost e=');readln(e);
clrscr;
x:=xn;
 
writeln('tabulirovanie finkcii y=Arth(x) dlyaя |x|<1');
writeln('na intervale [',xn:0:1,';',xk:0:1,'] s shagom ',dx:0:1);
writeln('-------------------------------');
writeln('|  x  |    S    |  arth   | N |');
writeln('-------------------------------');
while x<=xk+dx/2 do
 begin
  s:=0;
  si:=x;
  n:=0;
  i:=i+1;
  while abs(si)>e do
   begin
    dx:=dx+1;
    si:=si*x*x;
    s:=1/((s+si)*(2*i+1));
   end;
  writeln('|',x:4:1,' |',s:8:4,' |',  f(x):8:4,' |',  dx:3,'|');
  x:=x+dx;
 end;
writeln('---------------------------------');
readln;
end.