Open Science Hardware promises better science through greater collaboration, inclusion, freedom of innovation, customization and economy / frugality. A critical component to achieve these goals is the quality of documentation of designs published in academic journals and online. Are they available, complete and useful in order to reproduce and build on existing work? We present the results of a large scale, statistically representative, qualitative assessment of the quality of open scientific hardware publications, and associated characteristics such as venue of publication, evaluating their progress in time and regarding the establishment of standards for sharing open hardware designs. Our methodology is based on scraping a series of full-text searches for 'open hardware' as a term in Google Scholar, followed by metadata extraction using the API of the Web of Science (WoS) bibliographic information catalogue. We then proceed with a stratified sample of documents according to the dimensions that interest us, and download their full text for through inspection. The sampled documents are evaluated manually by a group of trained researchers in the area and graded according to a set of critical qualities related to optimal hardware design sharing. Our results shed a light on the evolving scenario of Open Science Hardware documentation practices, on the process of community learning around those practices, and the role of standards for adequately sharing hardware designs.