Since the start of nanoindentation in wood, cell walls have been supported by epoxy resin during the microtoming of surfaces. However, wood embedding in epoxy resin was initially developed to harvest thin sections destined for transmission electron microscopy. In nanoindentation, embedding can affect the identified mechanical properties and might be difficult for modified wood samples. For this reason, we propose a protocol that does not involve resin within the wood specimen, uses a more robust miller to achieve surfaces with root means squared roughness of around 35 nm, and can exclude invalid edge indents through a novel evaluation procedure based on fitting 2D histograms with a linear combination of bi-variate lognormal probability density functions. We examined twelve samples of Norway spruce (Picea abies) late wood and their counterparts soaked with a biological preservation pretreatment. A nanoindentation protocol on every sample led to a mean reduced modulus and mean hardness of around 12 % and 26 %, respectively, lower than those found in the literature for embedded Norway spruce. These results agree well with previous results found for Loblolly pine (Pinus taeda L.) when comparing embedded and non-embedded specimens.