The group of Harvard physics graduate students who devised a system last summer to beat the Massachusetts state lottery voted last week to dissolve their association, Greg Pollak, a group shareholder, said yesterday.

The system simply stopped making money, Claude Bernard '72, a shareholder and former graduate student, said Wednesday.

In its first ten weeks the group made a net profit of $400 on a $1200 investment--a 30 per cent rate of return. But, in the last four weeks profits dropped to under ten per cent and stayed there.

Several members of the group "felt it wasn't worth continuing the system," Bernard said.

Jonathan Sheiman and Edward Fahri, co-leaders of the group, refused to comment on the system's recent failure.

Patterns

The system was based on the students' observations that people playing the numbers tend to pick certain numbers more often than others.

The graduate students saw number picking patterns develop and bet on the numbers selected infrequently. The state pays a higher return on unpopular winning numbers.

Recently people began to bet on a broader range of numbers. The patterns no longer held and the system began to fail. Bernard said.

Bernard also said that the longer the group used the system, the less lucrative it became. After returns reached a peak profit point last month, the laws of probability took over and profits fell.