(a) Measured constraints on ﹩| \cos i| ﹩ and m₂ resulting from TEMPO fits for two different parametrizations of the Shapiro gravitational propagation delay within the context of general relativity. The Damour & Deruelle (1986) s and r parameters map directly onto the displayed ﹩| \cos i| \ [=+\sqrt{1-{s}^{2}}]﹩ and ﹩{m}_{2}\ [=(r/{T}_{\odot }){M}_{\odot }]﹩ axes, respectively; the black contours show joint 1, 2, and ﹩3\sigma \ ({\rm{\Delta }}{\chi }^{2}=2.3,6.2,11.8﹩) confidence limits on those quantities derived from a set of TEMPO fits to a large grid of (fixed) (﹩| \cos i| ,{m}_{2})﹩. The alternate Freire & Wex (2010) best-fit parameter constraints and their ±1σ limits are shown in green. Their fitted parameter, ς, transforms directly into the displayed ﹩| \cos i| ﹩ axis (see Equation (4)), whereas their h₃ parameter does not map uniquely onto either of the axes (see Equation (5)). The marginal distributions in (b) and (c) result from collapsing the resulting two-dimensional DD probability distribution onto the ﹩| \cos i| ﹩ and m₂ axes, respectively, in which the mean (solid black) and the 1σ bounds (gray region) are displayed, yielding ﹩| \cos i| =0.73\ (+0.05,-0.11)﹩ and ﹩{m}_{2}=1.95\ (+0.55,-0.71)\ {M}_{\odot }﹩ (68.3% confidence).