The linearity and the energy resolution of the calorimeter have been analysed by using electrons, pions and muons in an energy range from 1 GeV to 9.5 GeV. In order to study the calorimeter features, dedicated data acquisition runs were performed after removal of the TRD marble layers so that incoming particles do not interact before reaching the calorimeter.
Figures 5 and 6 display typical 3 GeV electron and pion events. At the left and right hand sides the TRD and calorimeter modules are shown respectively. The streamer tube system is also shown. As expected, after marble removal, both pions and electrons are recorded in the TRD as single tracks, while different shower patterns are visible in the calorimeter.
The calorimeter gain setting was performed by using a LED system, while the equalization was performed with throughgoing muons from cosmic rays and from the beam. In Fig. 7 the muon detection efficiency measured in the single calorimetric cell is reported.
Taking into account that a typical hadron calorimeter is 
interaction length thick, corresponding to 
cells for the
highly segmentated NO Ecalorimeter, 70% single cell efficiency
yields about 20 fired calorimeter cells, thus providing a high
efficiency for muon identification.
The observed detector performance have been compared with predictions
of a GEANT 3.21 based Monte Carlo program.
The simulation takes into account Birks saturation in the
scintillating fibers, Gaussian fluctuations of about 15% around the
measured fiber attenuation length, detector thresholds and non
linearities of readout electronics.
According to the measured light yield for a minimum ionising
particle, a mean value of 20 p.e./MeV with Poisson fluctuations on
each PMT is used.
The gain of the PMTs is set to 
, with 30% gaussian
fluctuations, coming from both quantum conversion efficiency and
electron multiplication.