Context : A large fraction of the carbon-enhanced extremely metal-poor halo giants ( [ Fe/H ] < ~ { } -2.5 ) are also strongly enriched in neutron-capture elements from the s -process ( CEMP-s stars ) .
The conventional explanation for the properties of these stars is mass transfer from a nearby binary companion on the Asymptotic Giant Branch ( AGB ) .
This scenario leads to a number of testable predictions in terms of the properties of the putative binary system and the resulting abundance pattern .
Some among the CEMP stars further exhibit overabundances in r-process elements on top of the s-process enrichment , and are tagged CEMP-rs stars .
Although the nucleosynthesis process responsible for such a mixed abundance pattern is still debated , these stars seem to belong to binary systems as do CEMP-s stars .

Aims : Our aim is to present and analyse in detail our comprehensive data set of systematic radial-velocity measurements and high-resolution spectroscopy of the CEMP star HE 0017+0055 .

Methods : Our precise radial-velocity monitoring of HE 0017+0055 over 2940 days ( 8 yr ) with the Nordic Optical Telescope and Mercator telescopes exhibits variability with a period of 384 d and amplitude of 540 \pm 27 m s ^ { -1 } , superimposed on a nearly linear long-term decline of \sim 1 m s ^ { -1 } day ^ { -1 } .
High-resolution HERMES/Mercator and Keck/HIRES spectra have been used to derive elemental abundances using 1-D LTE MARCS models .
A metallicity of [ Fe/H ] \sim - 2.4 is found , along with s-process overabundances on the order of 2 dex ( with the exception of [ Y/Fe ] \sim + 0.5 ) , and most notably overabundances of r-process elements like Sm , Eu , Dy , and Er in the range 0.9 – 2.0 dex .
With [ Ba/Fe ] > 1.9 dex and [ Eu/Fe ] = 2.3 dex , HE 0017+0055 is a CEMP-rs star .
The derived atmospheric parameters and abundances are used to infer HE 0017+0055 evolutionary status from a comparison with evolutionary tracks .

Results : HE 0017+0055 appears to be a giant star below the tip of the red giant branch ( RGB ) .
The s-process pollution must therefore originate from mass transfer from a companion formerly on the AGB , now a carbon-oxygen white dwarf ( WD ) .
If the 384 d velocity variations are attributed to the WD companion , its orbit must be seen almost face-on , with i \sim 2.3 ^ { \circ } , because the mass function is very small : f ( M _ { 1 } ,M _ { 2 } ) = ( 6.1 \pm 1.1 ) \times 10 ^ { -6 } M _ { \odot } .
Alternatively , the WD orbital motion could be responsible for the long-term velocity variations , with a period of several decades .
The 384 d variations should then be attributed either to a low-mass inner companion ( perhaps a brown dwarf , depending on the orbital inclination ) , or to stellar pulsations .
The latter possibility is made likely by the fact that similar low-amplitude velocity variations , with periods close to 1 yr , have been reported for other CEMP stars in a companion paper ( Jorissen et al. , 2015 ) .
Moreover , Kiss & Bedding ( 2003 ) have shown that Wood ’ s period-luminosity sequence D extends below the RGB tip , corresponding to periods of about 400 d , and is associated with velocity variations .
A definite conclusion about the origin of the 384 d velocity variations should however await the detection of synchronous low-amplitude photometric variations .

Conclusions :