Investigating the emotional impact of historical music, e.g. music of the 19th century, is a complex challenge since the subjects that listened to this music and their emotions are forever gone. As a result, asking them for their experiences is not possible anymore and we need other means to gain insights into the expressive quality of music of this century. In this vision paper, we describe a pattern-based method called MUSE4Music to quantitatively find similarities in different pieces of music. The reconstruction of musical patterns will allow us to draw conclusions from erratic documents that go far beyond the single pieces they are referring to.