Imagine a treasure trove of ancient knowledge locked away in a language few understand today. That's the challenge with Classical Arabic, a language rich with history, philosophy, and scientific advancements from a golden age, yet largely inaccessible to modern readers. Existing translation tools struggle, often mistaking Classical Arabic for its modern counterpart. Enter ATHAR, a new dataset poised to unlock these linguistic secrets. ATHAR, meaning "legacy" in Arabic, is a collection of 66,000 expertly translated passages from key Classical Arabic texts. These works, ranging from al-Tabari's historical chronicles to Ibn Sina's medical encyclopedia, represent a diverse tapestry of topics like science, medicine, philosophy, and culture. Unlike existing datasets that focus heavily on religious texts, ATHAR offers a broader perspective, providing valuable context and insights into the evolution of Arabic language and thought. Creating ATHAR wasn't without its challenges. Researchers had to meticulously clean and align the data, ensuring accuracy and consistency. They also had to address the problem of flipped text within the source material, where Arabic and English were sometimes mislabeled, requiring careful parsing and sorting. Tests with leading large language models (LLMs) like GPT-4 and Llama revealed the impact of this new dataset. While zero-shot performance (translation without prior examples) was limited, providing a few examples significantly improved accuracy, demonstrating the power of context. Fine-tuning the LLMs on ATHAR further boosted performance, particularly using the LoRA method which efficiently updates specific model parameters. The ATHAR dataset stands as a testament to the potential of focused data collection and careful curation. Not only does it provide a benchmark for evaluating current LLM capabilities, but it also paves the way for building more sophisticated and culturally nuanced translation systems. As ATHAR expands to include even more texts and topics, it promises to further bridge the linguistic divide, giving access to a rich intellectual heritage that has been largely hidden until now. This opens exciting possibilities for historians, researchers, and anyone curious about the fascinating world of Classical Arabic literature.