Spectral line observations are of vital importance in astrophysics. Hydrogen recombination lines are a uniquely powerful tracer of ionised gas in the Galaxy, and of star formation and gas kinematics in other galaxies. Improvements in receiver performance at GHz frequencies have generated renewed interest in recombination lines from other galaxies. The 76m Lovell Telescope should be sensitive enough to detect recombination lines from new galaxies in 10 - 100 hours. Jodrell Bank Observatory has been without a dedicated spectral line backend for more than 15 years. The advancement of digital technology means a step-change in bandwidth and resolution capability is now possible. A ROACH2 board was configured as a spectrometer with up to 15 kHz resolution in two polarisations across 512 MHz bandwidth and up to 62.5 kHz resolution across 2 GHz bandwidth in one polarisation. Signal transport systems were developed to allow observation of signals from the Mark II and Lovell telescopes by the spectrometer. The 2 GHz primary system was found to be dominated by correlated noise, introduced by the e-MERLIN frequency conversion system. Initial observations of the H96a-103a recombination line were found to be dominated by the band shape, which will limit calibration accuracy. The telescope was commissioned for use in spectral line observations. The C-band receiver showed good performance. A knee frequency of 17 mHz in a 250 kHz channel was found, scaling with bandwidth as expected. The telescope pointing accuracy was found to be a limiting factor, with offsets up to 50% of the half-power beam width observed. Nevertheless, hourly corrections demonstrated that sources could be tracked with 10% incremental offsets required. A one-hour observation of the calibrator NGC 7027 indicated 6% increase in the effective system noise, indicating long observations are well-represented by the radiometer equation. The RFI environment at Jodrell Bank was examined. C-band was found to be largely available for observation. The range of H96a-103a RRLs were reliably observable. L-band is densely populated with billion-Jansky signals. The protected band was observed over a four-week period using the Lovell Telescope. A strong correlation is found between human activity and the presence of RFI in the protected band, indicating that even where parts of L-band appear free from coherent RFI, sensitive total-power observations must take place at night. The future of RRL studies from Jodrell Bank is in C-band Galactic plane surveys using Mark II Telescope. Extragalactic RRL work is unfeasible. The Galactic anti-centre is sparsely observed. The Mark II Telescope is capable of a fully-sampled survey of 10 - 20 square degrees with millikelvin sensitivity in around one-hundred days' observing time.