The reduced transition probability B(E2;0 + gs → 2 + 1 ) for 28 S was obtained experimentally using Coulomb excitation at 53 MeV/nucleon. The resultant B(E2) value 181(31) e 2 fm 4 is smaller than the expectation based on empirical B(E2) systematics. The double ratio |M n /M p |/(N/Z) of the 0 + gs → 2 + 1 transition in 28 S was determined to be 1.9(2) by evaluating the M n value from the known B(E2) value of the mirror nucleus 28 Mg, showing the hindrance of proton collectivity relative to that of neutrons. These results indicate the emergence of the magic number Z = 16 in the |T z | = 2 nucleus 28 S. shown. They are associated with nuclear collectivity, which is enhanced, for instance, in the neutron-rich N = 20 nucleus 32 Mg caused by disappearance of the magic number [9,10].The new neutron magic number N = 16 has been confirmed experimentally for 27 Na (|T z | = 5/2) and more neutron-rich isotones [6-8, 11, 12]. Its appearance can be theoretically interpreted as a result of a large gap between the neutron d 3/2 and s 1/2 orbitals caused by the low binding energy [6] and/or the spin-isospin dependent part of the residual nucleon-nucleon interaction [13]. In analogy to the magic number N = 16, the proton magic number Z = 16 must also exist in proton-rich nuclei. However, it has not been identified experimentally in the proton-rich sulfur isotopes. The present Letter reports on a study of the magic number Z = 16 at the most proton-rich even-even isotope 28 S with |T z | = 2 through a measurement of the reduced transition probability B(E2;0 + gs → 2 + 1 ). The B(E2) value is directly related to the amount of quadrupole collectivity of protons.The relative contribution of the proton-and neutron-collectivities can be evaluated using the ratio of the neutron transition matrix element to the proton one (the M n /M p ratio) for An array of 160 NaI(Tl) scintillator crystals, DALI2 [24], was placed around the target to measure de-excitation γ rays from ejectiles. The measured full energy peak efficiency was 30 % at 0.662 MeV, in agreement with a Monte-Carlo simulation made by the GEANT4 code, and the energy resolution was 9.5 % (FWHM). The full-energy-peak efficiency for 1.5 MeV γ rays emitted from the ejectile with the velocity of 0.32c was evaluated to be 16% by the Monte-Carlo simulation.The scattering angle, energy loss (∆E), and total energy (E) of the ejectiles from the at the corners for the first two layers, and a 3 × 3 matrix for the third and fourth layers.The silicon detectors in the four layers had an effective area of 50 × 50 mm 2 and a thickness of 500, 500, 325, and 500 µm, respectively. The detectors in the first and second layers had 5-mm-wide strip electrodes on one side to determine the hit position of the ejectiles. The ∆E-E method was employed to identify 28 S. The mass number resolution for sulfur isotopes was 0.35 (1σ). The angle of the ejectile was obtained from the hit position on the telescope and the beam angle and position on the target measured by the PPACs. The scattering angle res...