Differences between 10Be records from Greenland and Antarctica over the last 100 years have led to different conclusions about past changes in solar activity. The reasons for this disagreement remain unresolved. We analyze a seasonally resolved 10Be record from a firn core (NEEM ice core project) in Northwestern Greenland for 1887-2002. By comparing the NEEM data to 10Be data from the NGRIP and Dye3 ice cores, we find that the Dye3 data after 1958 are significantly lower. These low values lead to a normalization problem in solar reconstructions when connecting 10Be variations to modern observations. Excluding these data strongly reduces the differences between solar reconstructions over the last 2000 years based on Greenland and Antarctic 1...