Keyword density penalty threshold

Is the threshold about 10% perhaps?

[Wizard]

Today I found one of my pages at #5 for certain query. But it wasn't exacly the page I intended for this query!

I designed a page to describe blue-golden widgets - with "blue-golden widgets" in title, h1, anchor text and some outgoing links, and of course in text content. This page has several links to subsections of the website about certain blue-golden widgets. I also put an outbound to official page about blue-golden widgets to let users find more information.

I also designed a page about the author of blue-golden widgets, with his name in title and h1, but also with links to blue-golden widgets and many occurences of "blue-golden widgets" in text.

And it is the second one on #5!

They are both related, however the 1st one is more related and provides more complete information. What can be wrong with this more related page? As checked with Webmaster World keyword density checker, first page has 'blue-golden widgets" density almost 13%, while the second one - a bit above 8%.

I didn't stuff keywords, this density was natural - the only way to lower this density would be to write more content with smaller density, as I cannot remove the keywords from where they are.

But what's most important - IS there a penalty for too big keyword density? A penalty, that lowers your position at certain query only?

The second page doesn't rank at the term it _is_ about - the name of the author - and his name has density about 13% too. So it may be another example of this penalty.

And my whole "golden widgets" site also does strange things - in search for "golden widgets" the pages that link to my are above my page, while the only occurence of "golden widgets" is in link to my page. I thought it was because they still have better PR, but maybe it's the penalty - "golden widgets" has density almost 12% on main page. So it could explain it - maybe the penalty lowers the position of my page.

We can expect there IS a penalty for too big density, but what would you say about the idea, that threshold is 10%? Or perhaps 8%?

[Chard]

I am sure there is probably a threshold, but I would have thought your pages were not excessive. My homepage used to have a k/w density of 23% for my main keywords, which used to worry me, though we ranked OK on G. I added some very useful and informative text to the page, to dilute the density, as, like you, I felt that I couldn't remove any of the keywords. With this text, the k/w density dropped to 1% - I had overdone it! However, it made absolutely no difference to my G positions. (Though I dropped on Yahoo!)I have since removed a lot of this text to its own page, linked from an introductory paragraph on the homepage, so that I could hopefully find a happy medium. Density rose to about 7% - made absolutely no difference to G, but I went back up on Yahoo. So that was the whole range from 1% to 23% and I never moved one slot up or down on G. I feel that the whole excercise achieved precisely nothing!
Like so many other things with G, I think k/w density is just one small part of their algo and if your site is borderline on some of the other factors, I suppose k/w density could trigger something, but exactly where that threshold is, I don't know. I have seen a recommendation on an authoritative? site, stating ideal k/w density of body text should be between 5% and 20%

[Wizard]

I have some pages with 40% keyword density, however it's mostly in img alt attributes, and they rank in top 10, so high density doesn't always trigger penalty or filter. But these observations I described in previous post made me wonder - I'll keep looking for the reason.

Perhaps Google doesn't use any thresholds for single factors, but always matches the coincidences of many factors. Maybe high density is ok, unless the phrase with high density doesn't appear in the URL? I'll keep researching.