Though hydrocarbons are widely viewed as being thermally unstable in nature, cracking to gas in reservoirs, there has been very little work to address the chemical feasibility of the concept. The present effort provides a test of this view by developing from laboratory experiments criteria for recognizing thermally altered hydrocarbons and then applying the criteria to two condensates that are believed to have been exposed to temperatures of 175-200 degrees C for 30 million years. The observation that the two samples exhibit none of the characteristics expected for thermal alteration products suggests that hydrocarbons are far more stable in nature than is generally recognized. The stability of hydrocarbons in nature means that an alternative mechanism must be devised to explain the presence of gas in deep reservoirs. A new mechanism is proposed which is expected to yield the appropriate isotopic and molecular distribution of products, operates in the source rock, uses established chemical pathways, and does not require any exotic catalysts.