With a resolution of 4.4 x 10(-3) cm(-1), we were able to identify in this range the very weak nu(2) + 2 upsilon(4)(0 )(A(1)) component near 1180 cm(-1), the upsilon(3) + upsilon(4) band around 1205 cm(-1) with its (A(1) + A(2)) and E very intermixed components, and the upsilon(1) + upsilon(4) (E) band centered at 1238 cm(-1). Three thousand six hundred transitions belonging to the (upsilon(3) + upsilon(4), upsilon(1) + upsilon(4)) interacting system were fitted together with a model taking into account l(2, 2) interactions inside upsilon(1) + upsilon(4)( )and between (A(1 )+ A(2)) and E components of upsilon(3) + upsilon(4), the l-vibrational resonance inside upsilon(3) + upsilon(4)(A(1) + A(2)), and the Coriolis interactions between upsilon(1) + upsilon(4) and upsilon(3) + upsilon(4 )(A(1) + A(2)) On one hand and between upsilon(1) + upsilon(4) and upsilon(3) + upsilon(4) (E) on the other. Four available MW transitions were also included in the fit. A rms of 0.76 x 10(-3) cm(-1) was obtained with 34 free parameters among 38. Normally the Fermi resonance, which links upsilon(3) to upsilon(2) + upsilon(4) with a coupling term W-234 = 2.86 cm must connect each component of upsilon(3) + upsilon(4) with each component of upsilon(2 )+ 2 upsilon(4). But since we have only little experiment information about the weak upsilon(2) + 2 upsilon(4)(0) component (120 assigned lines) and none about the dark upsilon(2 )+ 2 upsilon(4)(+/-2 )component It was not possible to introduce this resonance in the fit. However, the bandcenters' shifts were calculated since the basic coupling term W-234 and the anharmonic constant x(24) are well known. Therefore, according to this approximation, the very sensitive anharmonic constants x(34) and g(34) could be deduced. Of course the x(14) Fermi-independent constant, derived directly from the (upsilon(1) + upsilon(4))(0) bandcenter given by the fit, was certainly more accurate. (C) 1998 Academic Press.