Here's the NYT's graphs, which show not the expected rate per capita but the absolute number of perpetrators. The reader who doesn't stop to take this into account would assume that the black homicide rate is about twice the "white" rate. But, of course, there are far more white than black teens, so the black rate is actually much higher than just twice as bad.
First, Dr. Fox confirmed for me that what the New York Times referred to as number of homicides perpetrated by "whites" is actually the number for "whites plus most Hispanics." (The federal government is very careful about breaking out Hispanic statistics separately in almost all venues except crime rates. I wonder why?) This is an important point because it's likely that Hispanics now commit the majority of homicides in the younger age groups of the government's "whites plus Hispanics" aggregation.
Second, he sent me the per capita rates, from which I have calculated the ratio of homicide perpetrating rates. The following table lists the ratio of the murder rate for blacks to the murder rate for whites plus Hispanics. For example, in 2007, the black homicide rate among 14-17 year-old-males was 10.0 times the level of homicide rate for the aggregation of whites and Hispanics.