lm() for comparing groups

If the data are continuous, independent, the groups have equal variance and the residuals are normal, is the code below the correct test to do?

fit <- lm(mpg~gear, data=mtcars, subset=which(mtcars$gear==3|mtcars$gear==4))

Which of the following analyses can NOT be done using lm()?

If you were to compare eye colors of babies between different groups of babies, would you use lm()?

The biological effect (difference between the means) is represented by which parameter of the linear model?