In this work, we analyze the utility of two dimensional document maps for exploratory analysis of Polish case law. We start by comparing two methods of generating such visualizations. First is based on linear principal component analysis (PCA). Second makes use of the modern nonlinear t-Distributed Stochastic Neighbor Embedding method (t-SNE). We apply both PCA and t-SNE to a corpus of judgments from different courts in Poland. It emerges that t-SNE provides better, more interpretable results than PCA. As a next test, we apply t-SNE to randomly selected sample of common court judgments corresponding to different keywords. We show that t-SNE, in this case, reveals hidden topical structure of the documents related to keyword,,pension". In conclusion, we find that the t-SNE method could be a promising tool to facilitate the exploitative analysis of legal texts, e.g., by complementing search or browse functionality in legal databases.