> I just want to know if I'm specifying the graph wrong, or what,
> because this seems like really weird behavior:

> test_data<-
> data.frame(x=rnorm(100),y=rnorm(100),cat1=factor(rep(c(1,2),times=c(50,50)) ),cat2=rnorm(100))
> ggplot(test_data,aes(x=x,y=y)) +
> geom_point(aes(colour=cat1,size=cat2),alpha=1/2)

> versus

> ggplot(test_data,aes(x=x,y=y)) +
> geom_point(aes(colour=cat1,size=cat2))

> In the first version, the alpha value is clearly being applied to the
> _entire_ legend, even the size scale.  Am I placing the alpha
> parameter in the wrong place?

> Is there no way to force the alpha parameter to be applied only to the
> data points themselves, not the legend?

> I _think_ that the only way to do this is with grids and viewports.  You
> could capture a legend of  a plot without alpha set, create an
> appropriately dimensioned grid, print a plot (with alpha set and without
> a legend) to a viewport, print the captured legend to another viewport
> on the grid.  Maybe a better way will show up in the responses.

> Mark

> On 06/04/2010 05:58 PM,Joranwrote:

> > I've posted about a related issue before, which I was able to resolve
> > kind of, but it's popped up again with a slightly different graph.

> > I just want to know if I'm specifying the graph wrong, or what,
> > because this seems like really weird behavior:

> > test_data<-
> > data.frame(x=rnorm(100),y=rnorm(100),cat1=factor(rep(c(1,2),times=c(50,50)) ),cat2=rnorm(100))
> > ggplot(test_data,aes(x=x,y=y)) +
> > geom_point(aes(colour=cat1,size=cat2),alpha=1/2)

> > versus

> > ggplot(test_data,aes(x=x,y=y)) +
> > geom_point(aes(colour=cat1,size=cat2))

> > In the first version, the alpha value is clearly being applied to the
> > _entire_ legend, even the size scale.  Am I placing the alpha
> > parameter in the wrong place?

> > Is there no way to force the alpha parameter to be applied only to the
> > data points themselves, not the legend?