Randomly selecting a subset from a dataset with same instances from each class in WEKA

I'm having a dataset with 68 instances. Out of the 68 instances, 34 instances are in class A and 34 instances are in class B. I need to select 40 data points.

It should contain is 20 from class A and 20 from class B. The randomly selected data should have the same instances of data. This needs to be done using WEKA.

I was able to pick 40 random instances. But the number of instances from each class is not similar. Can someone please help me on how to do it using WEKA.

This is how I done up to now.

Shuffled the dataset randomly by, filter --> Randomize

selected 40 instances by, filter --> Resample.

But the number of instances received were no 20 from each class. How can I fix this.