We perform new explicit and regularization-independent calculations of < AV > amplitude in QED
1+1 and
amplitude in QED1+3 respectively to reinvestigate the anomaly problem. A kind of finite and unambiguous terms of exact scale behavior (in momentum space) is found to be responsible for the anomalies both in QED1+1 and QED1+1. They come from logarithmically divergent momentum integrals in contrast with the usual knowledge for triaugle anomaly in QED1+3. In QED1+1, this term is also responsible for photon mass generation. Some implications and possible relations with other aspects of anomaly are discussed.