Three programs, three different questions

JICA, 1988. The drilling arm of a three-phase Japanese government program that screened 30,000 km² and 25 sectors before choosing where to put a rig. 24 diamond holes for 1,204.8 m, average depth 50 m, a rig limit rather than a geological decision. Tundulu got the largest survey area of the three Chilwa carbonatite sectors: 7.5 km² and 243 ore samples, against 3.2 km² at Songwe Hill and 0.8 km² with no drilling at Kangankunde.

Optichem / Mota-Engil, 2014–15. 55 holes for 7,002 m, averaging 127 m, deepest 221 m. This crew was looking for phosphate. The rare earth intercepts are, in a real sense, a by-product of someone else's search.

AuKing, 2026. The first program designed to test the REE system on its own terms, and the first to go deep.

24 holes ended in mineralisation. Eight of those were still above 2 % TREO when they stopped. Use the "Flag holes open at depth" control on the map to pick them out.

The first assays

On 14 September 2026 AuKing reported assays from the first four RC holes. They validate the historical record and, in places, beat it.

26NH00591 m @ 1.51 % TREO

From 40 m, including 43 m @ 2.00 % from 77 m and 15 m @ 3.40 % from 90 m. Drilled beside historical ground to test whether the old record could be trusted.

26NH02980 m @ 1.54 % TREO

From surface, including 30 m @ 2.00 % from surface, 11 m @ 4.33 % from 2 m, 26 m @ 2.03 % from 53 m and 12 m @ 2.74 % from 56 m. Mineralised from the first metre.

26NH00420 m @ 2.06 % TREO

From 113 m, including 12 m @ 3.24 % from 113 m, and separately 10 m @ 2.86 % from 84 m. This is the twin of historical hole TU071.

26NH014step-out

Collared well south of Nathace to test a magnetic low. Several mineralised intervals, at lower grade, inside the anomaly. No headline intercept was quoted.

Set against the historical best of 41.1 m @ 3.68 % and 30 m @ 4.03 %, these are comparable grades over comparable widths, from holes sited to check the old work rather than to flatter it. Two of the three cored the mineralisation from surface or near it.

The whole record, metre by metre

Annexure B of that release publishes every one-metre sample. Below is the full downhole profile for each of the four holes: TREO grade as bars, the HREE+Y share of the basket as the green trace, and the announced intercepts bracketed down the right edge.

[downhole]

What the profile shows that a highlight box cannot is how the grade is distributed. 26NH029 is mineralised continuously from surface. 26NH005 carries a broad zone through the middle of the hole. 26NH004 has its best rock at the bottom, still going when the hole stopped at 145 m. And the green trace runs opposite to the bars in places: the heavy share tends to peak where TREO does not, which is the apatite signature rather than the carbonatite one.

Heavy rare earths, where they sit

Across the mineralised zones HREE+Y runs to roughly 8 % of TREO, with discrete intervals far above that. 26NH005 returned 8 m @ 16.09 % HREE+Y/TREO from 68 m, 7 m @ 12.75 % from 80 m and 6 m @ 8.88 % from 120 m. 26NH029 returned 42 m @ 11.46 % from 13 m, 4 m @ 9.45 % from 60 m and 12 m @ 8.54 % from 68 m. Over their full intercepts the two holes average 2.92 % and 4.53 % HREE+Y/TREO.

The elevated heavy grades track P₂O₅ across all three holes, which is consistent with the HREE-bearing apatite documented at Tundulu since the 1970s. That is the phosphate co-product and the heavy rare earth story turning out to be the same rock.

One correction. The HREE column printed for 26NH014 in Annexure B does not reconcile with the element data in the same rows; it runs about ten percentage points low. The profile above recomputes it from the reported oxides for that hole only. The other three holes reconcile exactly, as do all eleven announced TREO intercepts.

How the samples were assayed

Worth knowing, because rare earth assays are easy to get wrong. Some REE minerals are refractory, so a partial digestion undercounts them. AuKing used lithium-borate fusion, a total digestion, with preparation at Intertek's Kitwe laboratory in Zambia and the fusion and finish at Intertek Perth by ICP-OES and ICP-MS.

Certified reference materials (AMIS0356 and AMIS0939) inserted one in every twenty samples, and coarse blanks one in twenty.
QAQC samples made up 12.5 % of everything submitted, and the Competent Person reports the returns within acceptable limits.
Continuous 1 m sampling irrespective of lithology, riffle split to 2 to 4 kg, with a documented spear protocol where wet or clay-rich ground defeated the splitter.

What the 2026 rigs found

[hits total="31" hit="28" caption="One marker per completed RC hole across the finished 4,300 m program. Filled means carbonatite was logged. Logging is visual and interpretive; assays are what settle it."]

The hit rate held as the rigs moved away from Nathace. At the 27-hole mark it stood at 26 of 27, and every hole drilled into the Kamilala and Tundulu Hill step-outs came back with carbonatite, including multiple intrusions on both the western and eastern sectors of Tundulu Hill, 800 m east of the historical ground.

Drilling has now delineated carbonatite across three of the six interpreted complexes: Nathace, Kamilala and Tundulu Hill. Makhanga, Namuka and Chigwakwalu have still never been drill-tested. Cumulative logged carbonatite passed 2,000 m at the 27-hole mark and exceeded 2,100 m by the end of the RC phase.

Three things from the first half of the program are worth separating out.

The old data checks out. The first two holes, 26NH004 and 26NH005, were deliberately collared beside historical holes to test whether the inherited record could be trusted. No previous explorer had kept any RC chips or core, so there was nothing physical to re-log. Mafic dolerite intersections came in within a few metres of the depths previously reported.
Kamilala delivered on its first test. Two RC holes for 298 m into a target that had never been drilled, sited off the airborne magnetics rather than off historical workings, both intersecting carbonatite. That is the exploration model verifying itself.
The system is bigger than the old footprint. Carbonatite was logged well west and north-east of Nathace Hill, in ground historical drilling never touched.

Managing Director Paul Williams has framed the striking feature of the early drilling as carbonatite appearing across almost all holes, not just around Nathace, and Tundulu therefore having potential to be a much larger system than first thought.

An early field signal, heavily qualified. A video field update from senior geologist Derek Halford, posted to the company's social media rather than released to the ASX, reported portable XRF readings taken during drilling detecting rare earth elements in quantities of interest at Kamilala and Tundulu Hill.

Portable XRF is a screening tool used at the rig. It is indicative, not quantitative at these levels, and it is not a substitute for laboratory assay. Treat it as a pointer that the newly confirmed carbonatite may carry mineralisation rather than simply being favourable host rock, and nothing more than that.

Switch on the hole-trace layer to see collar azimuth and the surface projection of each angled hole.

[photo key="rig" alt="Drill rig on site at Tundulu with crew and stacked sample bags" caption="The rig on site, sample bags stacked at the collar. Every one of those bags is a few metres of the answer." credit="Project field photography"]

Down to 510.6 m

Diamond drilling opened with 26NH032, a hole already drilled by RC to 150 m. The plan was a 150 m diamond tail for 300 m total. Logging kept finding carbonatite past that point, so the hole was extended, and it finished at 510.62 m. That is 210 m beyond plan, on a rig that costs real money per metre and is not drilling another target while it sits there. Paying for those metres is a stronger signal than any wording in a highlights box.

How it finished matters more than how deep it went. Core photographed from roughly 505.2 m to the end of hole shows variably altered carbonatite and carbonatite breccias running to the bottom. The hole was stopped because the program had the data it needed at this stage, not because the geology ran out.

[scan src="https://bullseyeanalytics.net/wp-content/uploads/2026/08/tundulu-26nh032-core-505-510m-1.jpg" alt="Diamond drill core from 505.2 to 510.6 m in hole 26NH032 showing carbonatite and carbonatite breccias" caption="Core from about 505.2 to 510.6 m downhole in 26NH032, carbonatite continuing to end of hole." credit="AuKing, ASX release 17 Aug 2026"]

Every meaningful grade number in this project's history comes from a shallow hole, and many of those holes stopped while still in mineralisation. 26NH032 is the first to go looking underneath, and it found carbonatite for its entire length without reaching the bottom of the system. The company's own framing shifted accordingly: the surface footprint initially strengthened the district-scale interpretation, and the depth extension now indicates the complex may be significantly larger than previously interpreted. The model grew on both axes at once.

One detail to watch, and to not over-read. Red-orange alteration phases are visible in the core at around 449 m downhole. They remain subject to detailed logging and analytical confirmation. Alteration is not mineralisation, and visual observation is not assay.

Where the samples actually are

The assay pipeline crosses three countries, which is most of the reason it has been slow. The position as at the 17 August 2026 release:

1,270At Intertek Perth, being assayed

4,165Export approved, 7 permits

357Still with the MMRA

CoreSampling started, none reported

The first two RC batches are in Perth with priority processing requested for the first four drillholes. The bulk of the rest is at or in transit to Intertek's sample preparation facility in Zambia, with one permit covering 357 samples still with the Malawi regulator.

Diamond core is logged, photographed and cut lengthways, with half submitted for analysis and half retained in labelled trays. Sampling runs on nominal 1 m intervals, flexing between 0.5 m and 1.5 m to honour geological and mineralisation boundaries, and core stays under company supervision through to AuKing's core processing and storage facility at Zomba. Thompson Drilling is running the diamond rig on NQ-diameter core. None of this is glamorous, and all of it is what a defensible resource estimate is built from.

Best historical intercepts

AuKing cites three headline historical intercepts when describing the project: 41 m at 3.7 % TREO from 8 m, 35 m at 2.7 % TREO from 15 m, and 30 m at 4.03 % TREO from surface. The first and third are JMT-22 and TU014 below. The hole behind the second is not identified in the public summaries. The 2026 assays above now sit alongside these rather than beneath them.

[intercepts]

Surface sampling

Sample	TREO	Detail
7Y386	6.41 %	JICA 1988, Nathace Hill. Highest single surface sample
7Y196	6.07 %	JICA 1988, Tundulu Hill east flank, 5.6 % P₂O₅. Highest grade from a hill nobody had drilled
TUTR10	≤ 3.35 %	Trench, Oct 2024. 63 samples over 83 m; P₂O₅ to 27.5 %; HREO ~13 % of basket
Sat-1	3.10 %	Kamilala Hill satellite target
BC2B	2.56 %	DY6 reconnaissance, carbonatite
7Y380	2.06 %	JICA, carbonatite breccia
Tundu-1	1.80 %	DY6 reconnaissance, brecciated fenite

Of 85 rock chips assayed by JICA, 25 came back above 1 % TREO.

The 1989 historical estimate

From that drilling on the eastern slope of Nathace Hill, JICA published a preliminary estimate of about 0.6 Mt at 2.09 % REO across three ore bodies, plus roughly 0.5 Mt at 17.0 % P₂O₅. It is a keyhole number: one hillside, outcropping zones only, a 1 % REO cut-off and an assumed 50 m depth limit, and the report says plainly that it cannot be compared with operating mines. Songwe's equivalent 1988 estimate of 1.4 Mt has since become 21 Mt. The full context, with the scanned pages, is in the JICA 1987 panel.

The 2026 program

31 holesRC, complete 31 Jul 2026

4,300 mRC metres

28 / 31Holes with carbonatite

2,100 m+Cumulative logged carbonatite

510.6 m26NH032, ended in carbonatite

400.4 m26NH023, ended in carbonatite

Both diamond holes were planned to 300 m and both were extended because logging kept finding carbonatite. Core from them is earmarked for metallurgical and mineralogical testwork as well as assay.

Samples travel Malawi to Zambia to Intertek in Perth. Management has publicly acknowledged shareholder frustration at how long that chain has taken, while arguing it has still moved faster than the regional norm. The current position is set out below.

Surveyed collars

Collar positions for the first 21 RC holes were published in the announcement of 2 July 2026 and are plotted here at their surveyed coordinates, converted from WGS 84 UTM Zone 36 South. Holes prefixed 26NH sit in the Nathace Hill area, 26KH at Kamilala. The remaining ten holes of the 31-hole RC program are not yet published at collar level.

[collars]

Grid is WGS 84 UTM Zone 36 South. Reported depths are downhole; true widths are not yet known. Everything else on this map, including geological outlines and historical collars, remains schematic and compiled from public releases rather than survey data.