function cin (x)
  c december 1980 edition.  w. fullerton, bell labs.
        dimension cincs(11)
        external  csevl, inits, r1mach
  c
  c series for cin  on the interval  0.00000e+00 to  1.60000e+01
  c                                        with weighted error   7.33e-17
  c                                         log weighted error  16.14
  c                               significant figures required  15.42
  c                                    decimal places required  16.66
  c
        data cin cs(  1) /     0.3707450175 090968874 e0/
        data cin cs(  2) /    -0.0589357489 636444683 e0/
        data cin cs(  3) /     0.0053818964 211356912 e0/
        data cin cs(  4) /    -0.0002986005 284196214 e0/
        data cin cs(  5) /     0.0000109557 257532162 e0/
        data cin cs(  6) /    -0.0000002840 545487735 e0/
        data cin cs(  7) /     0.0000000054 697399488 e0/
        data cin cs(  8) /    -0.0000000000 812418746 e0/
        data cin cs(  9) /     0.0000000000 009586859 e0/
        data cin cs( 10) /    -0.0000000000 000092027 e0/
        data cin cs( 11) /     0.0000000000 000000733 e0/
  c
        data eul / 0.5772156649 0153286 e0 /
        data ncin, xmin /0, 0.0/
  c
        if (ncin.ne.0) go to 10
        ncin = inits (cincs, 11, 0.1*r1mach(3))
        xmin = sqrt (r1mach(1))
  c
   10   cin = 0.0
        absx = abs(x)
        if (x.ne.0.0 .and. absx.le.xmin) call seteru (
       1  38hcin     x so small that cin underflows, 38, 1, 0)
        if (absx.le.xmin) return
  c
        if (x.gt.4.0) go to 20
        cin = x*x*csevl ((x*x-8.0)*.125, cincs, ncin)
        return
  c
   20   call r9sifg (x, f, g)
        sinx = sin (absx)
        call erroff
        cin = -f*sinx + g*cos(absx) + alog(absx) + eul
  c
        return
        end