The ultra-compact dipping source \object{XB 1916-053} has an orbital period of close to 50 min and a companion star with a very low mass (less than 0.1 M$_{\odot}$). The orbital period derivative of the source was estimated to be $1.5(3) \times 10^{-11}$ s/s through analysing the delays associated with the dip arrival times obtained from observations spanning 25 years, from 1978 to 2002. The known orbital period derivative is extremely large and can be explained by invoking an extreme, non-conservative mass transfer rate that is not easily justifiable. We extended the analysed data from 1978 to 2014, by spanning 37 years, to verify whether a larger sample of data can be fitted with a quadratic term or a different scenario has to be considered. We obtained 27 delays associated with the dip arrival times from data covering 37 years and used different models to fit the time delays with respect to a constant period model.We find that the quadratic form alone does not fit the data. The data are well fitted using a sinusoidal term plus a quadratic function or, alternatively, with a series of sinusoidal terms that can be associated with a modulation of the dip arrival times due to the presence of a third body that has an elliptical orbit. We infer that for a conservative mass transfer scenario the modulation of the delays can be explained by invoking the presence of a third body with mass between 0.10-0.14 M$_{\odot}$, orbital period around the X-ray binary system of close to 51 yr and an eccentricity of $0.28 \pm 0.15$. In a non-conservative mass transfer scenario we estimate that the fraction of matter yielded by the degenerate companion star and accreted onto the neutron star is $\beta = 0.08$, the neutron star mass is $\ge 2.2$ M$_{\odot}$, and the companion star mass is 0.028 M$_{\odot}$. (Abridged)

We analyse all available X-ray observations of X1822-371 made with XMM-Newton, Chandra, Suzaku and INTEGRAL satellites. The observations were not simultaneous. The Suzaku and INTEGRAL broad band energy coverage allows us to constrain the spectral shape of the continuum emission well. We use the model already proposed for this source, consisting of a Comptonised component absorbed by interstellar matter and partially absorbed by local neutral matter, and we added a Gaussian feature in absorption at $\sim 0.7$ keV. This addition significantly improves the fit and flattens the residuals between 0.6 and 0.8 keV. We interpret the Gaussian feature in absorption as a cyclotron resonant scattering feature (CRSF) produced close to the neutron star surface and derive the magnetic field strength at the surface of the neutron star, $(8.8 \pm 0.3) \times 10^{10}$ G for a radius of 10 km. We derive the pulse period in the EPIC-pn data to be 0.5928850(6) s and estimate that the spin period derivative of X1822-371 is $(-2.55 \pm 0.03) \times 10^{-12}$ s/s using all available pulse period measurements. Assuming that the intrinsic luminosity of X1822-371is at the Eddington limit and using the values of spin period and spin period derivative of the source, we constrain the neutron star and companion star masses. We find the neutron star and the companion star masses to be $1.69 \pm 0.13$ M$_{\odot}$ and $0.46 \pm 0.02$ M$_{\odot}$, respectively, for a neutron star radius of 10 km.In a self-consistent scenario in which X1822-371 is spinning-up and accretes at the Eddington limit, we estimate that the magnetic field of the neutron star is $(8.8 \pm 0.3) \times 10^{10}$ G for a neutron star radius of 10 km. If our interpretation is correct, the Gaussian absorption feature near 0.7 keV is the very first detection of a CRSF below 1 keV in a LMXB. (abridged)

The bursting pulsar, GRO J1744-28, went again in outburst after $\sim$18 years of quiescence in mid-January 2014. We studied the broad-band, persistent, X-ray spectrum using X-ray data from a XMM-Newton observation, performed almost at the peak of the outburst, and from a close INTEGRAL observation, performed 3 days later, thus covering the 1.3-70.0 keV band. The spectrum shows a complex continuum shape that cannot be modelled with standard high-mass X-ray pulsar models, nor by two-components models. We observe broadband and peaked residuals from 4 to 15 keV, and we propose a self-consistent interpretation of these residuals, assuming they are produced by cyclotron absorption features and by a moderately smeared, highly ionized, reflection component. We identify the cyclotron fundamental at $\sim$ 4.7 keV, with hints for two possible harmonics at 10.4 keV and 15.8 keV. The position of the cyclotron fundamental allows an estimate for the pulsar magnetic field of (5.27 $\pm$ 0.06) $\times$ 10$^{11}$ G, if the feature is produced at its surface. From the dynamical and relativistic smearing of the disk reflected component, we obtain a lower limit estimate for the truncated accretion disk inner radius, ($\gtrsim$ 100 R$_g$), and for the inclination angle (18$^{\circ}$-48$^{\circ}$). We also detect the presence of a softer thermal component, that we associate with the emission from an accretion disk truncated at a distance from the pulsar of 50-115 R$_g$. From these estimates, we derive the magneto-spheric radius for disk accretion to be $\sim$ 0.2 times the classical Alfv\'en radius for radial accretion.