Whatever or wherever it is leaking, it's spreading.
Despite TEPCO's attempt to tell us the cesium contamination in the observation hole No.1-2 was from contaminated dirt particles (or other residues), that still doesn't account for higher all-beta and very high tritium.
Now, the sample taken from another observation hole, No.3, on July 11 suddenly shows higher all-beta amount.
From TEPCO's email notice for the press, 7/12/2013:
Groundwater observation hole No.3
・７月11日採取分 Sample collected on July 11：
セシウム134 Cs-134 1.9 Bq/L
セシウム137 Cs-137 4.8 Bq/L
全ベータ all-beta 1,400 Bq/L
・７月４日採取分（お知らせ済み） Sample collected on July 4 (already published)：
セシウム134 Cs-134 1.5 Bq/L
セシウム137 Cs-137 2.8 Bq/L
全ベータ all-beta ＮＤ（18 Bq/L）
Here's the diagram showing where these holes are located. They are along the seawall, east of the turbine building.
From Nuclear Regulatory Agency (secretariat of NRA) document, 7/10/2013 (original diagram from TEPCO, 6/26/2013):
In the earlier version of the diagram by TEPCO (6/16/2013), the observation hole No.3 is right near the location where a leak of highly contaminated water from Reactor 3 (remember it is a MOX-fuel reactor) was found on May 11, 2011.
The Tokyo University researcher who has written papers on measuring radioactive materials of Fukushima origin, including the one about neptunium-239 discovery in Iitate-mura tweeted an interesting idea on how to find out whether the contamination is coming from any of the reactors:
How about mixing a small amount of radionuclides that are not present in the reactors in the water to cool the reactors, and seeing if these radionuclides are detected in the observation holes?