Timing and scintillations of the millisecond pulsar 1937+214

Pulse shapes, arrival times, and interstellar scintillations of the 1.56 ms pulsar are analyzed at frequencies from 0.32 to 1.4 GHz using data obtained from 1983 April to 1985 December. The two pulse components (main pulse and interpulse) are different in shape, and their separation is weakly frequency dependent, decreasing from 173°.6 to 173°.0 of pulse phase between 0.43 and 1.4 GHz. Pulse shapes at the lowest frequencies show broadening caused by scattering in the interstellar medium that is consistent with the intensity scintillations that are also seen. The occurrence of fast scintillations of the pulsar intensity demonstrates that the broadening is caused by multipath scattering rather than by angular wandering of a single ray path. We test the precision to which the measured pulse phase represents the true rotational phase of the pulsar. Intrinsic phase jitter of individual pulses (∼15 μs) causes time of arrival errors in sums of N pulses that scale roughly as N^-1/2. At low frequencies, the largest time of arrival errors are due to interstellar scintillations. Scintillation-induced frequency structure changes on time scales ∼1 minute, introducing arrival time errors of a few microseconds. On time scales of months and longer, the measured pulse phase varies in a wavelength-dependent manner. The long-term, wavelength-dependent time of arrival variations, if interpreted as dispersion measure changes, are δDM ≈ 0.003 pc cm^-3 over 1000 days. However, it is unclear whether the wavelength dependent variations are due solely to DM variations. At two epochs, the variations at three frequencies show the ν² scaling expected from dispersion measure variations. However, comparison of DMs calculated from 1.4 to 2.4 GHz data (published by Rawley et al. in 1988) with those calculated from 0.43-1.4 GHz data are inconsistent: the first set is systematically larger than the second set of DMs. The bias may indicate that (1) arrival times are perturbed by changes in pulse shape with frequency; (2) there are additional contributions from interstellar scattering, including angle of arrival effects that contribute ∼ν² and ∼ν⁴ perturbations to arrival times; (3) the volume of interstellar scattering material that is sampled is a function of frequency, owing to the scaling of the scattering diameter ∝ν^2.2; and (4) there is nonsimultaneous emission of different frequencies toward Earth due to a variation in altitude of emission combined with rotational aberration, reference frame dragging, gravitational bending of rays, and magnetic field line distortion. Further exploration of these possibilities will require additional measurements at many frequencies between 0.3 and 3 GHz. Time series of scintillation parameters are consistent with scattering in the interstellar medium from electron density irregularities with a spectrum (wavenumber)^-α with α = 3.55 ± 0.11. The Kolmogorov spectrum (α = 11/3) is consistent with these results. The range of length scales encompassed by the spectrum is at least 10¹¹-10¹⁴ cm with the lower limit probably extending down to 10⁹ cm or less. The distribution of scattering material along the line of sight appears to be nearly uniform. Refraction from large-scale irregularities in the ISM evidently produces angular wandering of the pulsar image that is much less than the diffractive broadening of the image.