Results are presented of experimental investigations of the beam pattern (BP) of the RATAN-600 radio telescope observing the bright discrete sources. The measurements were carried out in the range from 1.4 to 49 cm wavelength at the elevations of sources from 10 to 90 degrees. The main beam of the BP was measured to a level of 0.5 - 3% of the BP maximum. The aberration curves were measured that describe the fall-down of signal in observations with transverse shifts of the horn from the antenna focus. The measurements of the BP made it possible to check the accuracy of calculation and reveal important effects which influence on the structure of BP far out of its main axes. A comparison was made of the drift scans of the Moon with the convolutions of the two-dimensional computed BP with the uniform disk of angular dimensions of the Moon, the root-mean-square error of setting the main mirror elements is estimated. It is equal to 0.55 mm. It is show that the new computation of the BP [1], with takes account of the diffraction effects and the finite size of the RATAN-600 main mirror ring have a better fit to the data of measurements than the early calculations [2].

The BP of the North sector of the RATAN-600 radio telescope was measured in three sets of observations (2001 February-March, 2001 Oktober, 2002 April). Sources were observed in the mode of transit of a source across the immovable BP of the telescope. The flux densities of the sources were above 3 Jy. The higt sensitivity radiometric complex of the feed-cabin No.1 was used for the measurements. The sensitivity of its radiometers is from 2.5 mK to 20 mK. With the effective area of about 1000 m2 a high signal-to-noise ratio was realized in one observation even in the cross-sections far from the central one. In the process of observations the drift scans of point sourse  ( PKS0521-36,  PKS1830-21,  3C161,  3C454.3, 3C121,  2005+403,  3C84) with an elevation H were registered. Transits of sources were observed across the horizontal cross-sections of the BP which differed in elevation from the central horizontal cross-section by H.

A comparison of the computed and experimental relations  Fmax(H), as well as comparison of drift scans of point sources across different horizontal cross-sections of the BP with the corresponding computed cross-sections were made. Under a good state of the antenna, the concidence of the experimental and computed curves is reached at the level 2-3% of the BP maximum Fmax(0) in the central cross-section. In a good state of the antenna the root-mean-square error of setting the elements of the main mirror in radius and elevation is not worse than 2-2.5 precise scale graduations of the synchro.

Individual cross-sections of the BP were simulated in the presence of errors in setting the panels in elevation and radial coordinates. It is show how the structure of the BP changes at distant cross-sections with changes of the value and character of distribution of these errors over the aperture of the main mirror.

As can be seen from the curves shown in Fig.4, precence of panels with large elevation errors lead to a considerable change in the BP structure. These changes are the more pronounced the shorter is the wavelength. On condition that the number of panels with great errors is rather large, a general shift of the BP in elevation may occur if errors of one sign predominate. The radial errors also have an effect on the BP structure, but considerably weaker than errors in positions of the panels in elevation. Any errors in setting of the reflecting elements of the calculed position cause not only distortion of the BP structure but also a fall-down in gain of the antenna.

A modeling of transit of an extended source across the computed two-dimensional BP of RATAN-600 and a comparison of the results of modeling with actual records of transit of the Moon across the BP were made. Convolutions of the computed BP with the uniform disk of angular dimensions of the Moon. The background scattering brought by occasional errors of setting the panels was simulated at short wavelengths ( 1.38, 2.7, 3.9 cm). It was specified by the two-dimensional Gaussian, the halfwidths of which are equal in the horizontal and vertical plane the width of panel and the effective vertical size of the panel correspondingly. Occacional errors do not affect the shape of the drift scan at wavelengths > 7.6 cm. At these wavelengths the experimental drift scans of the Moon and the convolutions of the two-dimensional BP were compared with the disk of homogeneous brightness of the Moon's angular dimensions. In all the records, but for those with dominating receiver's noises, the experimental and calculated curves are almost identical.

Relations between the maximum values of the BP in the central cross-section Fmax and the transverse displacement of the receiving horn from the antenna focus (the so called aberration curves) were measured using the reference point sources. It can be seen that at elevations < 80o computed curves are in good agreement with the results of measurements. The discrepancy in plotting the computed and experimental curves at elevation 88o with very large displacements of the horn from the focus is likely to be associated with incorrect setting of the horn at wavelength 1.38 cm along the focal line of the secondary mirror. Probably, additional measurements are needed to refine the results obtained.