A Fullo worked in a Fullonica, i.e., a place to launder Roman clothes. From my calendar: In ancient Rome, clothes were laundered in human urine, collected from public latrines. Urine contains ammonia, a natrual whitener. The clothes were then rinsed in water to remove the urine smell. The process was quite effective for cleaning clothes but no much fun for the workers at the laundry, or fullonica, who had to spend the day standing in vats of urine stomping on pee-soaked garments.
Edited to answer Rajini Rao’s question about the bile staining the Roman linens.
The clothes were then hung on a vessel of basket-work (viminea cavea), under which sulphur was placed in order to whiten the cloth; for the ancient fullers appear to have known that many colours were destroyed by the volatile steam of sulphur (Apul. Met. IX. p208, Bipont; Plin. H. N. XXXV.50, 57; Pollux, VII.41). A fine white earth, called Cimolian by Pliny, was often rubbed into the cloth to increase its whiteness (Theophr. Char. 10; Plaut. Aulul. IV.9.6; Plin. H. N. XXXV.57).
I have to dig out my photos of Ostia.
Hopefully appropriate for ScienceSunday if not maybe #scienceeveryday
Here’s a news article, In cancer science, many ‘discoveries’ don’t hold up, about an oncology researcher trying to replicate some preclinical studies before moving forward with potential drug development. (thanks to a post via Branimir Vasilić http://goo.gl/wJyMx)
The news article summarizes a commentary in the journal Nature, titled, Drug development: Raise standards for preclinical cancer research.
Notice the difference in the titles? Here’s a similar discussion where Rajini Rao points out that the news article is titled, Eggs unlimited: an extraordinary tale of scientific discovery vs. Potential Egg Stem Cells Reignite Debate in the journal Science. Similar discussion here: http://goo.gl/Yq1ls
I want to focus on the oncology debate since I do cancer research. However, the comments from the article and me are relevant to many areas of research.
Drug development: Raise standards for preclinical cancer research
C. Glenn Begley & Lee M. Ellis
Nature 483, 531–533 (29 March 2012) doi:10.1038/483531a
Published online 28 March 2012
Here are 5 reasons why oncology research might not be replicated
Endpoints
As the authors point out, endpoints in cancer research can be less quantitative compared to say statin research where cholesterol level is the endpoint. In cancer studies sometimes tumor size is an endpoint. As an imaging person, my field very frequently frowns on this, as a drug can cause tumor swelling, i.e., increase in size, while actually causing tumor cell death. Not everyone has access to expensive imaging equipment or the skills to utilize many imaging modalities. So a lot of cancer drug researchers rely on caliper measurements of the tumor even though most would acknowledge that a tumor is rarely a perfect sphere where one only needs to measure the diameter.
Cutting edge
The authors suggest that some of the irreproducible results could be due to publications that were cutting edge, i.e., a researcher found something completely new or unexpected and published quickly. Also some technology might not be available to Amgen that was used in one of the publications. For example oxygen imaging is available in maybe 3-4 labs in the world.
Competition
Although this may sound terrible to the general public, there have been cases where researchers have omitted a key ingredient or step on a method in order to keep a competitive advantage.
Narrow scope
Begley and Ellis state that the robustness of some results were checked. For example, a publication might get phenomenal results with a particular tumor cell-line or model. When Amgen tried to broaden the scope, e.g., trying a different cell-line or model, the “narrow” promising results turned out to be less robust.
Statistics
Another issue is improper statistics. Quite often scientist haven’t had enough statistical training or do not consult a statistician and therefore use an incorrect method or interpretation.
Conclusion
Interestingly, Begley mentions that the results do not use enough predictive biomarkers (an area of focus for my research which I hope to contribute a solution). The authors’ suggestion to try to show tumor models where there is a negative result is often not possible when a grant funds a particular cancer or model. I totally agree about the selective presentation aspect of their paper. Unfortunately, I don’t think it is uncommon for a publication to have a figure that is stated to be “representative” of all the data, when in fact it was carefully selected as the best example. As some commenters on the online version of this Nature article state, it’s interesting that Begley and Ellis do not list the publications they tried to replicate, thereby limiting the possibility to replicated their article. Transparency?
Edit: I want to be clear that I don’t condone some of these reasons for the lack of reproducible publications. I want to emphasize that there are some reasons why a drug company might not be able to replicate a publication and therefore, there is no need for Reuters or Yahoo news to say the sky is falling for scientist.
