A new technique for the spatial deconvolution of spectra is applied to near-IR ( 0.95 - 2.50 \mu ) NTT/SOFI spectra of the lensed , radio-quiet quasar HE 1104 - 1805 .
The continuum of the lensing galaxy is revealed between 1.5 \mu and 2.5 \mu .
Although the spectrum does not show strong emission features , it is used in combination with previous optical and IR photometry to infer a plausible redshift in the range 0.8 < z < 1.2 .
Modeling of the system shows that the lens is complex , probably composed of the red galaxy seen between the quasar images and a more extended component associated with a galaxy cluster with fairly low velocity dispersion ( \sim 575 km s ^ { -1 } ) .
Unless more constrains can be put on the mass distribution of the cluster , e.g .
from deep X-ray observations , HE 1104 - 1805 will not be a good system to determine H _ { 0 } .
We stress that multiply imaged quasars with known time delays might prove more useful as tools for detecting dark mass in distant lenses than for determining cosmological parameters .

The spectra of the two lensed images of the source are of great interest .
They show no trace of reddening at the redshift of the lens nor at the redshift of the source .
This supports the hypothesis of an elliptical lens .
Additionally , the difference between the spectrum of the brightest component and that of a scaled version of the faintest component is a featureless continuum .
Broad and narrow emission lines , including the FeII features , are perfectly subtracted .
The very good quality of our spectrum makes it possible to fit precisely the optical Fe II feature , taking into account the underlying continuum over a wide wavelength range .
HE 1104 - 1805 can be classified as a weak Fe II emitter .
Finally , the slope of the continuum in the brightest image is steeper than the continuum in the faintest image and supports the finding by Wisotzki et al .
( 1993 ) that the brightest image is microlensed .
This is particularly interesting in view of the new source reconstruction methods from multiwavelength photometric monitoring .
While HE 1104-1805 does not seem the best target for determining cosmological parameters , it is probably the second most interesting object after Q 2237+0305 ( the Einstein cross ) , in terms of microlensing .