Characteristics and sources of ambient particle elements in urban Beijing were studied by hourly observations in two size fractions (PM₁₀ and PM_2.5) during November and December 2017 using an online multi-element analyzer. The reconstructed oxide concentrations of 24 elements (from Al to Pb) comprise an appreciable fraction of PM₁₀ and PM_2.5, accounting for 37% and 17%, respectively on average. We demonstrate the benefit of using high-time-resolution chemical speciation data in achieving robust source apportionment of the total elemental PM₁₀ (PM_10el) and elemental PM_2.5 (PM_2.5el) mass using positive matrix factorization (PMF). Biomass burning, coal combustion, secondary sulfate, industry, non-exhaust traffic and dust were identified in both size fractions (with varying relative concentrations), which accounted on average for 4%, 12%, 5%, 2%, 14%, and 63%, respectively to the total PM_10el, and 14%, 35%, 21%, 6%, 12% and 12%, respectively to the total PM_2.5el. Biomass burning and coal combustion exhibited higher concentrations during haze episodes of the heating season. In contrast, secondary sulfate and industry contributed more to haze episodes during the non-heating season. The fractional contribution of dust was mostly high during clean days, while the fractional non-exhaust traffic emission contribution was similar throughout the measurement period. The non-exhaust traffic emissions contributed locally, while the remaining sources were dominated by neighboring areas. Furthermore, trajectory analysis showed that the origin of the industrial sources roughly agreed with the locations of the main point sources. Overall, this work provides detailed information on the characteristics of the elements during different haze events during heating and non-heating seasons.