Help! Type1 P-value seems significant, but Type3 p-value seems INsignificant

Hello! :) Does anyone can help me finding out if this predictor is significant or insignificant?

Here is my SAS code, response has two value, 0 and 1

#1) proc genmod ;
class spring summer fall winter;
model response=spring summer fall winter/dist=binomial link=logit type1 type3;
run;

#2) proc genmod ;
class spring summer fall winter;
model response=spring fall summer winter/dist=binomial link=logit type1 type3;
run;

#1 shows that only 'spring' is significant in the model, under Type1 and Type3
#2 shows that both 'spring' and 'fall' are significant under Type1, but only 'spring' is significant under Type3.
So the variable 'fall' seems insignificant if the variables order like spring,fall,summer and winter
but 'fall' seems significant if they order like spring,summer,fall and winter.:confused:

I googled, and here is my conclusion.
For the first case, 'summer', which is insignificant, is added to the model before 'Fall', so 'fall' seems insignificant.
For the second case, only 'spring' is added to the model before 'fall', so 'fall' seems significant.
But I don't really know if 'fall' is truly significant in that model or not...please help!:wave:
Is 'fall' significant in that model?