function ci (x)
c december 1980 edition, w. fullerton, bell labs.
      dimension cics(13)
      external alog, cos, csevl, inits, r1mach, sin, sqrt
c
c series for ci   on the interval  0.00000e+00 to  1.60000e+01
c                                        with weighted error   1.94e-18
c                                         log weighted error  17.71
c                               significant figures required  17.74
c                                    decimal places required  18.27
c
      data ci  cs(  1) /    -0.3400428185 6055363156 e0/
      data ci  cs(  2) /    -1.0330216640 1177456807 e0/
      data ci  cs(  3) /     0.1938822265 9917082877 e0/
      data ci  cs(  4) /    -0.0191826043 6019865894 e0/
      data ci  cs(  5) /     0.0011078925 2584784967 e0/
      data ci  cs(  6) /    -0.0000415723 4558247209 e0/
      data ci  cs(  7) /     0.0000010927 8524300229 e0/
      data ci  cs(  8) /    -0.0000000212 3285954183 e0/
      data ci  cs(  9) /     0.0000000003 1733482164 e0/
      data ci  cs( 10) /    -0.0000000000 0376141548 e0/
      data ci  cs( 11) /     0.0000000000 0003622653 e0/
      data ci  cs( 12) /    -0.0000000000 0000028912 e0/
      data ci  cs( 13) /     0.0000000000 0000000194 e0/
c
      data nci, xsml /0, 0.0/
c
      if (ntci.ne.0) go to 10
      nci = inits (cics, 13, 0.1*r1mach(3))
      xsml = sqrt (r1mach(3))
c
 10   if (x.le.0.0) call seteru (
     1  17hci      x is le 0, 17, 1, 2)
c
      if (x.gt.4.0) go to 20
      y = -1.0
      if (x.gt.xsml) y = (x*x-8.0)*0.125
      ci = alog(x) - 0.5 + csevl (y, cics, nci)
      return
c
 20   call r9sifg (x, f, g)
      sinx = sin (x)
      call erroff
      ci = f*sinx - g*cos(x)
c
      return
      end