First poll after Likud’s primary leaves Netanyahu’s bloc where it has been all summer

The Times of Israel’s Hebrew sister site Zman Yisrael published a seat projection on 21 August showing Yashar, Gadi Eisenkot’s party, slipping to 23 seats while Likud held at 21, with Naftali Bennett’s Together gaining at Eisenkot’s expense and the overall deadlock intact. The Times of Israel’s own summary framed the result as suggesting that Likud’s primary election that week “didn’t have any significant impact.” That reading is the outlet’s, not a pollster’s, and it should be attributed as such. Two caveats travel with the number: we hold the section index rather than the article permalink, and no sample size was published in what reached us. Earlier Zman Yisrael waves logged in the public polling tables were conducted with samples of 500, which is the scale at which a two-seat move is not a finding.

Set against the series, the summer has been static at the top and volatile beneath it. On 5 June a poll cited by the Times of Israel had Likud 23, Together 21, Yashar 19. By 12 June Zman Yisrael had Yashar and Together tied on 21, both behind Likud; by 18 June Yashar had drawn level with Likud and Together had fallen to 19. Midgam on 4 August gave Likud and Yashar 23 each and Together 14. Two days later Yashar led 23 to 22. A Channel 12 poll in mid-August put Yashar first on 24 to Likud’s 22; a Kan 11 wave reversed it, Likud 24 to Yashar 23, with 41 percent naming Netanyahu the better leader against 34 percent for Eisenkot. On 17 August Haaretz reported the two tied on 23, the coalition parties on 52, and the Arab parties unchanged at 11.

The structural figure is steadier than any of them. The Centre for Israel Education, aggregating late-June and July surveys, found the Likud–ultra-Orthodox–far-right bloc taking only 49 to 52 seats across that period, well short of the 61 needed, while Netanyahu himself stayed personally competitive and Likud stayed at or near the top of the field. The centre is a US education non-profit rather than a polling house, and its own summary notes Reuters reading the accumulated July numbers, on 29 July, as pointing to a clear Netanyahu defeat. Haaretz maintains a running tracker averaging the Channel 11, 12 and 13 polls, with Channel 14, Channel 15 and Maariv displayed separately.

Assessment: What is actually moving is the traffic between Bennett and Eisenkot: one pool of voters shuttling between two vehicles, which reads as drama and changes no arithmetic. Likud’s own number is the stable object in the data, and that is why a leadership primary was never going to shift it. Treat any single-poll lead change as noise until three houses agree. The number that decides October is the bloc line, and the opposition’s majority in most projections exists only if it counts the eleven Arab-party seats it has not said it will use. The variable that could reset a frozen race is not the slate. It is an Iran emergency of the kind the same papers led with this morning.