Duets Blog asked about the use of texture as a trademark.
I'm in favor of registering anything (compatible with public policy) that can put the public on reasonable notice that the thing is used a trademark. By reasonable notice, I mean there is consistent reproducibility in the product or service and adequacy of the description in the registration.
As a former manufacturing engineer, I am familiar with the problems of reproducible manufacturability and in reliable measurement (description). Fortunately, technology has reduced both to essentially 'no problem' for many fields of manufacture.
As such, the question of reasonable notice now shifts to adequacy of the description for public notice. While an embossed sample apparently sufficed for OLD PARR (Ecuador?), current electronic searches (via TESS, the Internet, etc.) require a textual description with a reasonable pictorial facsimile for non-word visual marks. This is not to prohibit embossed masters for registration, but the public needs reasonable notice without having to seek and buy every competitive product in the field. Some tactile marks could meet the textual description with a reasonable pictorial facsimile criterion. More complicated tactile marks might require dimensional (surface finish) descriptions, the measurement of which is now standardized and is therefore adequate for putting the public on notice. Holography would also be a good descriptive viewer with proper technology (but not on my monitor as yet).
Moving on to scent, non-functional single description scents have had minor success for registration as a trademark, but a description as a mix of scents failed (http://www.caslon.com.au/trademarkprofile11.htm). From an engineering and forensic point of view, a description based on gas-liquid chromatography results would likely have reproducibility, adequacy of the description, and be electronically searchable. Multiple blends of scents might be registered if the GLC description truly described the scent as reproduced with pure chemicals. Even if there were some difficulty to electronic searches, the difficulty would likely not be more than patent lawyers (like me) currently work with for prior art searches (and yes, it leaves a lot to be desired).
This leaves the fifth sense, taste, but taste analysis sufficient for reproducible manufacturing is still a skilled talent outside of an engineered metric. As evident by the description of fruit flavor and oak tone in wine, taste relies considerably on smell. The wide range of genetic ability to taste various compounds combined with age and culture related considerations makes an adequate description likely impossible at any stage of technology.