We present experimental results on statistics of polymer orientation angles relatively to shear plane and tumbling times in shear flow with thermal noise. Strong deviation of probability distribution functions (PDF) of these parameters from Gaussian was observed and a good accord with theory was found. The scaling relations of PDF widths for both angles as a function of the control parameter W i are verified and compared with numerics. An universal exponential PDF tail for the tumbling times and its predicted scaling with W i are also tested experimentally against numerics.PACS numbers: 23.23.+x, 56.65.Dy Dynamics and statistics of a single polymer molecule in stationary as well as in random flows have recently attracted attention of both experimentalists [1][2][3][4] and theorists [5][6][7][8][9][10]. Stretching dynamics and statistics and coil-stretch transition in these flows were investigated in detail. Besides, another remarkable effect was first observed in a shear flow [2], namely large fluctuations in a polymer elongation due to end-to-end aperiodic tumbling (see Fig.1D).Recently rather extensive theoretical [11,12] and numerical [13,14] efforts were conducted with the goal to understand the statistics of the angular orientation of a polymer molecule and of the tumbling time in a random velocity field with a mean shear. Shear flow with a thermal noise was considered there as a particular case. A role of thermal noise on a solid rod tumbling in a shear flow was first considered in Ref. [15]. Then the role of the Brownian fluctuations in the polymer dynamics and statistics was studied in numerical simulations [7], where, however, only the power spectral density and the statistics of polymer extension in a shear flow were investigated. An average polymer extension and angular orientation were theoretically considered also in Ref. [8], where the results of calculations were compared with light scattering measurements [16,17].In this Letter we concentrate on statistics of angular orientation and tumbling and scaling relations of their characteristics for a single DNA molecule in a shear flow, when thermal fluctuations are the main cause for tumbling.It is well known that a probability distribution function (PDF) of the end-to-end vector R for a polymer described by a dumbbell model with a linear relaxation in a simple shear is Gaussian [7]. Nevertheless, PDFs of the polymer extension, | R| ≡ R, as well as polymer angular orientation are strongly non-Gaussian due to anisotropy introduced by the shear. A functional form of the PDF of the polymer extension in a shear flow was first identified experimentally [2] and then explained theoretically and numerically [7].The main result of the recent theory [11,12] is the prediction of the intermittent (non-Gaussian) statistics of polymer angular orientation in a shear flow. Moreover, it was also shown that intermediate asymptotic of the angular polymer statistics bears some universal features independent of the nature of polymer random excitation [11,13,14]. The uni...