Generally speaking, one-way anova is pretty robust to unequal variance, especially when the group size are equal.
For you case, the major concern here is the sample size of group 4 (>50) is too small compared to other groups. If I were you, I would combine group 3 (41-50) and group 4 (>50), run the one-way ANOVA again and don't care about the equal or unequal variance any more.

If you really care about the unequal variance and don't want to combine group 3 and 4, the following are two choices:
          1) In SPSS one-way ANOVA "Options", check "Brown-Forsythe" or "Welch". There two procedures are supposed to deal with one-way ANOVA with unequal variance
          2)Nonparametric Kruskal-Wallis ANOVA