A precise measurement of the energy is essential for the study of ultra-high energy cosmic rays. Basically, two types of detectors are used for this purpose: Surface arrays which sample the shower tail: this method records the lateral development of the shower of secondary particles using an array of particle detectors and Fluorescence detectors which record the longitudinal development of the shower and observe the atmospheric fluorescence induced by charged particles in the shower. The second method is currently the most precise one to estimate the energy of cosmic rays. Fluorescence detectors provide a measurement of primary cosmic ray energy which is relatively model independent, as the fluorescence intensity is proportional to the electromagnetic energy released by the shower into the atmosphere. The main source of systematic uncertainties for energy estimation comes from the limited accuracy in the measurement of the air-fluorescence yield. A new type of absolute measurement of the nitrogen fluorescence yield in the air will be performed at LAL using 3 items which will yield an very accurate precision in all conditions of pressure, temperature, and pollutants. A 5 MeV electron beam will be provided by the new electron accelerator PHIL at LAL. As the fluorescence yield is proportional to the energy loss of the electrons, the contribution of secondary electrons (deltas) to the signal is much more important than the contribution of the primary electrons. It has therefore been chosen to use an integrating sphere, the basic property of which being that the probability to detect light is independent from where the light is produced inside the sphere. An output device on this sphere will be equipped with a set of optical fibers driving the fluorescence light to a Jobin-Yvon spectrometer equipped with an LN2 cooled CCD. The fluorescence spectrum in the 300-400 nm range will be accurately measured in steps of 0.1 nm resolution. A PMT equipped with a BG3 filter (the same as on JEM-EUSO) will be set on the sphere to measure the integrated yield. The sphere will be monitored by a NIST photo-diode, and will be surrounded by a spherical envelope to create a temperature controlled chamber (a Dewar). With this setup it will be possible to vary the temperature from -60degC to +40degC and the pressure from 1 to 0.01 atm. The expected precision of the yield should be better than 5%.