The Elmal guards the stead heroquest is considered one of the most difficult quests to pull off, and the best way to do it is not obvious. I'm hoping this guide may help some people who play the game without reloading.

Also, I don't seem to be able to post images in the discussion boards so this is an easier place to share my findings.

King of Dragon pass is a random game, so the only way to find out how something affects a heroquest is by playing through it a lot and recording the outcomes for different parameters. In this case, these parameters are the heroquester's stats. The other variables were kept as stable as possible.

Since this is a hard heroquest, a player will likely arrange some very favorable circumstances to perform it under. Because of this, I made sure that my circumstances were favorable as well. Below you can see the stats of my clan that were used.


The clan was made to contain about 770 adults (Farmers, weaponthanes, nobles crafters and hunters). A chalana arroy shrine for curing kept the number of sick relatively stable. A ridiculous amount of magic was stored up to minimize the influence of the heroquester being unable to enter the realm of the gods. The clan used is an Elmali peace clan, and it is part of a tribe with 8 other clans, which generally provided 160-170 tribe members to help with the heroquest. The heroquest was performed in the same year, which was reloaded to each time the heroquest was failed or succeded at. Three points of clan magic were allocated to quests and nothing else. The heroquest was performed in a year where the clan ring no longer indicated that they had heroquested recently. In order to prevent the magic stat from having a double effect by increasing both the heroquester's skill and the highest magic skill on the clan ring, the clan leader was trained to have a heroic magic skill. She can be seen on the bottom right of this section.

Heroquests were performed with the following responses: The responses to the eater of skin, author of sores and maker of bad growth were chosen in no particular order, except for starting by thrusting the spear into the eye of the eater of skin initially. Elmal was given tears between fights. The old love was told "Orlanth did not steal my pride. He awakened it.", and the oracle was told "I would not be loyal to Orlanth if I believed him so easily defeated.". The teller of lies was offered elmals soul in the form of a fine stallion, and promised to serve chaos each run. Each run was done for the purpose of strengthening clan leaders, just in case that affected anything. The heroquest was not performed on years that were good for heroquesting or years in which general good fortune was predicted. Lastly, in a stroke of luck, an event where a delegation of the clan's men would complain about the clan leader would trigger at the start of each year, so the effects of events could be kept constant as well by picking the 'do nothing' option at the start of each year.

The first experiment was based on a suspicion I had: The oracle tries to convince Elmal to be a leader, and Elmal is tempted into becoming a king when the old love section is failed. So, perhaps these sections are actually negatively affected by leadership?

To test this, I rerolled my clan until I got an Elmal worshipper with high magic and poor leadership (Terasarin), and then trained a second one(Asborn) to have heroic leadership, magic, custom, bargaining and combat. It is not possible to raise these stats without increasing leadership, and they were maxed out since improving them(at the cost of increasing leadership) might offset a reduction in performance brought about by high leadership. Animals was not trained, since most male clan members can only train their animals skill by doing the elmal guards the stead heroquest, and obviously it would be silly to use the heroquest to train for itself. Coincedently, these two heroquesters happen to have an equal animal skill, so the difference will not affect their performance. Each heroquester replaced the Elmal godtalker on the ring before performing Elmal guards the stead, since ring members are better at several things and may be better at performing heroquests as well. Replacing the elmal worshipper meant that each run was done with 7 different representatives on the ring.

Although i had initially planned to perform 100 heroquests per heroquester, this turned out to be very time consuming. After I performed 70 heroquests with asborn, I settled for doing 50 heroquests on subsequent godtalkers.

The results of the experiment are shown in the table below:


In case you're interested, here is the breakdown of the causes of failures:

We'll be limiting our analysis to the regular sucesses and failures of the heroquester. Asborn and Terasarin have sucess rates of 0.6 and 0.62 respectively. No significant difference can be detected between the two, p=0.41. Therefore, the difference in stats between the two does not seem to matter. Asborn's heroic leadership is presumably not adversely affecting him.

The effect of leadership was not the only I set out to test. Since Elmal's quest is the only quest that can raise the animals skill on a normal heroquester, it seems like a reasonable assumption that animals might help the quester. I had therefore prepared a barntar worshipper with an heroic animals skill to perform the heroquest. Unlike Elmal worshippers, Barntar worshippers can have their animals skill raised by being on the ring while the lankhor mhy heroquest is performed to make the clan leaders wiser. If Argrand's heroic animals skill does not help him, he would be expected to fail badly due to him not being an Elmal worshipper and his terrible stats. Argrand also replaced the Elmal worshipper on the clan ring. Since there was no barntar worshipper on the clan ring, the clan retained seven different representatives.


Argrand has a success rate of 0.52, which is not significantly different from Asborn's(p = 0.26).

It seems that Argrand's animals skill might be helping. But it is also possible that the sheer size of the clan and tribe are maxing out the chance of success. Therefore, I also tried to use a worshipper whose stats were all around poor, Hankorl the Lankhor Mhy worshipper who was already on the ring. This meant that these experiments were done without first reorganising the clan. Since Hankorl is also different from Argrand by having no combat skill to speak of, tests were also done on the trickster who had high animals skill but no significant skill in combat. Although Garstal's animals skill is listed as renowned in the image on the right, his skill actually fluctuated between excellent and renowned depending on the run.

Initially Garstal performed worse than Hankorl. Since this was a very odd result, I did another 50 runs on both to double check. These are the results from that:

Here is a breakdown of the failures, for anyone interested.

There is no significant difference between Garstal and Hankorl (p=0.28), who have success rates of 0.46 and 0.42, respectively. Both Garstal and Hankorl have a significantly lower success rate than Asborn: p<0.05 for Garstal and p<0.01 for Hankorl. Given these results, it seems that animals does not have a significant effect on performance, but combat does. However, since Argrand's combat and magic skills are merely 'good', it seems that these skills do not need to be especially high under favorable circumstances.

While I was doing my experiments, I became curious about the results of the other responses to the oracle and old love. 'This oracle is false' , 'I can't abandon these people', and 'I loved you once' did not seem better than the normal responses. However, telling the old love that she was a chaos creature in disguise would sometimes result in the heroquester thinking he was right, and the 'I can be a king, but that does not mean Orlanth is dead' option seemed to produce a more positive result than simply reaffirming the quester's loyalty to Orlanth. With these responses, I tested Asborn and Terasarin again. Note: After seeing good results with Asborn, I did not bother to put Terasarin on the ring for tests, being off may have adversely affected him.


Aborn and Terasarin both perform significantly better with these responses than when they simply follow the myth. p < 0.05 for Asborn and p<0.01 for Terasarin. What is perhaps more telling is the distribution of failures:


As you can see, having the heroquester accept kinghood from the oracle always resulted in success for both heroquesters.

To check if the improvement can only be attributed to the response to the oracle, the success rate of specifically the old love section can be used. Successes with the old love will result in either success of the quest, or failure at the oracle. This produces the following stats for both heroquesters:


Asborn has a significantly better performance when calling out the old love as a chaos creature(p<0.05). Terasarin performs about the same(p=0.40). Perhaps Asborn's higher stats make the chaos creature option more effective for him. Or perhaps the measured change in Asborn's performance is merely a fluke.

It seems that magic and combat are the most important stats for this heroquest, but under favorable circumstances you do not need much of either. That doesn't mean you should be training your heroquesters in these stats, though: Even though Asborn and Terasarin outperformed Hankorl with normal responses, Hankorl would still have a better chance of performing the heroquest than Terasarin this way if he does it twice(66% chance vs 62%). So it's probably not worth it to train up a single heroquester.

Even Hankorl barely ever failed to knit himself together, so it does not seem worth it to sacrifice anything other than tears so long as your clan is big.

None of my heroquesters failed at tearing the teller of lies apart from the inside, and accepting the crown from the oracle never failed. If you do end up failing either of these responses, please share a screenshot. It would be neat to see. It's possible that accepting the crown will lead to a bad event later down the line. If it does, I'll add that to this guide. But even if it does, at least you will have gotten the quest out of the way as a requirement for beating the game.

Edit: I have now managed to fail accepting the crown. My temple to orlanth was destroyed and the other clans seemed to dislike mine. This was an Ernalda clan performing the quest, perhaps accepting the crown only works for Elmal clans?

I hope I'm using statistics theory correctly here. If I'm making any mistakes, feel free to let me know.

• Magic and combat are the only relevant skills. Leadership does not hurt your heroquester's chances.
  
• It's better to try the quest with many godtalkers than to train up one really good one, at least so long as your clan and tribe are big.
  
• It's enough to give tears between fights, at least when your clan and tribe are big.
  
• At the old love section, if your skills besides magic and combat are high, choose "You're just a chaos creature in disguise." before "Orlanth did not steal my pride. He awakened it.". Otherwise, just choose "Orlanth did not steal my pride. He awakened it.".
  
• At the oracle section, pick "I can be a king, but that does not mean Orlanth is dead."

Hope this helps!