This diploma thesis is focused on the plasma surface treatment of historical glass from the 18th and 19th centuries by low temperature electrical discharges in solutions of sodium chloride and potassium carbonate and finding the most suitable settings of conditions for the surface cleaning. Stainless steel electrode and a specially designed electrode system with wolfram wire in the quartz glass capillary were used for the generation of discharge using an audio frequency power supply. Each line of samples was made from one piece of historical glass that was cut to smaller pieces. All cleaned samples were photographed before and after the cleaning so the possible changes of the cleaned area could be visually compared. Then the samples were analysed by LA-ICP-MS (line scanning of surface), where was analysed the cleaned area of samples, and values were compared to the analysis of the reference sample that was not cleaned. Examined isotopes of elements were selected on the basis of the supposed composition of glass, corrosion products, and soil at the place of discovery. Analyses were standardized by NIST 610. Acquired values were transferred to oxides. The most important oxides (Na2O, MgO, SiO2, P2O5 a K2O) were chosen for deciding the most effective cleaning settings. It was decided that the most effective setting for cleaning was the one where the biggest difference of values between sample and reference occurred. Four series of these solutions were compiled and one parameter was changed for each of them. Solutions and their conductivity, frequency of the power supply, and time of cleaning were chosen as changing values. Three samples of different times of cleaning were cleaned without interruption. The time of cleaning was split into intervals of 30 seconds of cleaning and 1 minute of non-action for another two samples of this series. In this way we were trying to find out if the following surface analysis will be influenced by the diffusion of the particles into the sample, or not. The frequency of power supply was recorded and its dissipated power was calculated for each measurement. Emission spectra of a series of different solution conductivity were measured before cleaning of samples. Measurement of OES was made with the ignition of discharge so the active species of plasma were shown in spectra. These species are probably participating in the cleaning process of glass. Emission spectra were also measured after cleaning to find out if values of active species were changed or unknown spectral lines appeared. These lines should be from dirt and corrosion products that were cleaned from the surface of the glass. It was found out that the most effective cleaning of sample 1 (series where the conductivity of the NaCl solution was changed) was done in a solution of conductivity 900 S/cm. The most effective cleaning of sample 4 and sample 7 (series where the conductivity of the K2CO3 solution was changed) was done in a solution of conductivity 600 S/cm. The most effective cleaning of sample 6 (series where the frequency was changed) was done at frequency = (15200 ± 30) Hz. The most effective cleaning of sample 5 (series of different cleaning times) lasted seven minutes without time delay. The future research it should be appropriate to try a combination of these most effective cleaning settings on the surface of more samples, so the finding of this thesis will be confirmed.