This document summarizes a study on using Word2Vec to analyze word embeddings over time in a historical Swedish newspaper corpus from 1749-1925. The study tracks how the meanings of 11 words change over time by training Word2Vec models on the corpus yearly and comparing the resulting word vectors. Preliminary results show word vectors become more stable as word frequency increases, and that on average words share meanings with 2-3 other words each year. Examples of shifting meanings for words like "woman", "politics", and "happy" are provided in Swedish. Future work involves handling OCR errors and finding diachronic word replacements and sense-based embeddings to better capture word sense changes over time.