We present early time high-resolution ( VLT/UVES ) and late time low-resolution ( VLT/FORS ) optical spectra of the normal type Ia supernova , SN 2001el . The high-resolution spectra were obtained 9 and 2 days before ( B-band ) maximum light . This was in order to allow the detection of narrow hydrogen and/or helium emission lines from the circumstellar medium of the supernova . No such lines were detected in our data . We therefore use these spectra together with photoionisation models to derive upper limits of 9 \times 10 ^ { -6 } ~ { } { M } _ { \odot } ~ { } { yr } ^ { -1 } and 5 \times 10 ^ { -5 } ~ { } { M } _ { \odot } ~ { } { yr } ^ { -1 } for the mass loss rate from the progenitor system of SN 2001el assuming velocities of 10  km s ^ { -1 }  and 50  km s ^ { -1 } , respectively , for a wind extending to outside at least a few \times 10 ^ { 15 } cm away from the supernova explosion site . So far , these are the best H \alpha based upper limits obtained for a type Ia supernova , and exclude a symbiotic star in the upper mass loss rate regime ( so called Mira type stars ) from being the progenitor of SN 2001el . The low-resolution spectrum was obtained in the nebular phase of the supernova , \sim 400 days after the maximum light , to search for any hydrogen rich gas originating from the supernova progenitor system . However , we see no signs of Balmer lines in our spectrum . Therefore , we model the late time spectra to derive an upper limit of \sim 0.03 M _ { \odot } for solar abundance material present at velocities lower than 1000  km s ^ { -1 }  within the supernova explosion site . According to numerical simulations of Marietta et al . ( 2000 ) this is less than the expected mass lost by a subgiant , red giant or a main-sequence secondary star at a small binary separation as a result of the SN explosion . Our data therefore exclude these scenarios as the progenitor of SN 2001el . Finally , we discuss the origin of high velocity Ca II lines previously observed in a few type Ia supernovae before the maximum light . We see both the Ca II IR triplet and the H \& K lines in our earliest ( - 9 days ) spectrum at a very high velocity of up to \sim 34 000 km s ^ { -1 } . The spectrum also shows a flat-bottomed Si II ‘ 6150 à ’ feature similar to the one previously observed in SN 1990N ( Leibundgut et al . 1991 ) at 14 days before maximum light . We compare these spectral features in SN 2001el to those observed in SN 1984A and SN 1990N at even higher velocities .